CHAPTER 1 INTRODUCTION 1.

Introduction to Digital Image Processing:
   Vision allows humans to perceive and understand the world surrounding us. Computer vision aims to duplicate the effect of human vision by electronically perceiving and understanding an image. Giving computers the ability to see is not an easy task - we live in a three dimensional (3D) world, and when computers try to analyze objects in 3D space, available visual sensors (e.g., TV cameras) usually give two dimensional (2D) images, and this projection to a lower number of dimensions incurs an enormous loss of information.    In order to simplify the task of computer vision understanding, two levels are usually distinguished; low-level image processing and high level image understanding. Usually very little knowledge about the content of images High level processing is based on knowledge, goals, and plans of how to achieve those goals. Artificial intelligence (AI) methods are used in many cases. High-level computer vision tries to imitate human recognition and the ability to make decisions according to the information contained in the image.   This course deals almost exclusively with low-level image processing, high level in which is a continuation of this course. Age processing is discussed in the course Image Analysis and Understanding, which is a continuation of this course.

1.1 History:
Many of the techniques of digital image processing, or digital picture processing as it was often called, were developed in the 1960s at the Jet Propulsion Laboratory, MIT, Bell Labs, University of Maryland, and few other places, with application to satellite imagery, wire photo standards conversion, medical imaging, videophone, character recognition, and photo enhancement. But the cost of processing was fairly high with the computing equipment of that era. In the 1970s, digital image processing proliferated, when cheaper computers Creating a film or electronic image of any picture or paper form. It is accomplished by scanning or photographing an object and turning it into a matrix of dots (bitmap), the meaning of which is unknown to the computer, only to the human

viewer. Scanned images of text may be encoded into computer data (ASCII or EBCDIC) with page recognition software (OCR).

1.2 Basic Concepts:
 

A signal is a function depending on some variable with physical meaning. Signals can be
o o o o

One-dimensional (e.g., dependent on time), Two-dimensional (e.g., images dependent on two co-ordinates in a plane), Three-dimensional (e.g., describing an object in space), Or higher dimensional.

1.3 Pattern recognition
Pattern recognition is a field within the area of machine learning. Alternatively, it can be defined as "the act of taking in raw data and taking an action based on the category of the data" . As such, it is a collection of methods for supervised learning. Pattern recognition aims to classify data (patterns) based on either a priori knowledge or on statistical information extracted from the patterns. The patterns to be classified are usually groups of measurements or observations, defining points in an appropriate multidimensional space. Are to represent, for example, color images consisting of three component colors.

1.4 Image functions:
 

The image can be modeled by a continuous function of two or three variables; Arguments are co-ordinates x, y in a plane, while if images change in time a third variable t might be added.

 

The image function values correspond to the brightness at image points. The function value can express other physical quantities as well (temperature, pressure distribution, distance from the observer, etc.).

The brightness integrates different optical quantities - using brightness as a basic quantity allows us to avoid the description of the very complicated process of image formation.

The image on the human eye retina or on a TV camera sensor is intrinsically 2D. We shall call such a 2D image bearing information about brightness points an intensity image.

  

The real world, which surrounds us, is intrinsically 3D. The 2D intensity image is the result of a perspective projection of the 3D scene. When 3D objects are mapped into the camera plane by perspective projection a lot of information disappears as such a transformation is not one-to-one.

 

Recognizing or reconstructing objects in a 3D scene from one image is an ill-posed problem. Recovering information lost by perspective projection is only one, mainly geometric, problem of computer vision.

The second problem is how to understand image brightness. The only information available in an intensity image is brightness of the appropriate pixel, which is dependent on a number of independent factors such as
o

Object surface reflectance properties (given by the surface material, microstructure and marking),

o O

Illumination properties, And object surface orientation with respect to a viewer and light source.

CHAPTER 2 2. DIGITAL IMAGE FORENSICS
Today's technology allows digital media to be altered and manipulated in ways that were impossible twenty years ago. We are feeling the impact of this technology in nearly every corner of our lives, from the courts to the media, politics, business, and science. As this technology continues to evolve it will become increasingly more important for the science of digital forensics to keep pace. This presentation will describe state of the art techniques in digital image forensics. Digital watermarking has been proposed as a means by which an image can be authenticated. This approach works by inserting at the time of recording an imperceptible digital code (a watermark) into the image. With the assumption that tampering will alter a watermark, an image can be authenticated by verifying that the extracted watermark is the same as that which was inserted. The major drawback of this approach is that a watermark must be inserted at precisely the time of recording, which limits this approach to specially equipped digital cameras. In contrast, recent advances in digital forensics operate in the absence of any watermark or specialized hardware. With the assumption that tampering disturbs certain underlying statistical properties of an image, these forensic techniques can detect specific forms of tampering. Air-brushing or re-touching can be detected by measuring deviations of the underlying color filter array correlations. Specifically, virtually all digital cameras record only a subset of all the pixels needed for a full-resolution color image. Instead, only a subset of the pixels is recorded by a color filter array (CFA) placed atop the digital sensor. The most frequently used CFA, the Bayer array, employs three color filters: red, green, and blue. Since only a single color sample is recorded at each pixel location, the other two color samples must be estimated from the neighboring samples in order to obtain a three-channel color image. The estimation of the missing color samples is referred to as CFA interpolation or demosaicking. In its simplest form, the missing pixels are filled in by spatially averaging the recorded values. Since the CFA is arranged in a periodic pattern, a periodic set of pixels will be precisely correlated to their neighbors according to the CFA interpolation algorithm. When an image is re-touched, it is likely that these correlations will be destroyed. As such, the presence or lack of these correlations can be used to authenticate an image, or expose it as a forgery. A digital composite of two people can be detected by measuring differences in the direction to the illuminating light sources from their faces and body. By making some initial simplifying assumptions about the light and the surface being illuminated, we can mathematically express how much light a surface should receive as a function of its position relative to the light. A surface that is

pictures. a region growing algorithm combines any significant number of neighboring blocks that are consistent with the cloning of an image region. If a digital watermark distorts the carrier signal in a way that it becomes perceivable. "Watermarking" is the process of hiding digital information in a carrier signal. Any inconsistencies in lighting can then be used as evidence of tampering. This form of tampering can be detected by first partitioning an image into small blocks. Once expressed in this form. the hidden information should. Digital watermarks may be used to verify the authenticity or integrity of the carrier signal or to show the identity of its owners. a fragile watermark would be applied. digital watermarks are only perceptible under certain conditions. and imperceptible anytime else. standard techniques can be used to determine the direction to the light source for any object or person in an image. Traditional Watermarks may be applied to visible media (like images or video). Instead. it is of no use.1. after using some algorithm. It is prominently used for tracing copyright infringements and for banknote authentication. Duplication or cloning is a simple and powerful form of manipulation used to remove objects or people from an image. a digital watermark does not change the size of the carrier signal.directly facing the light. texts or 3D models. will be brighter than a surface that is turned away from the light. their presence can then be used as evidence of tampering. . The blocks are then re-ordered so that they are placed a distance to each other that is proportional to the differences in their pixel colors. Since it is statistically unlikely to find identical and spatially coherent regions in an image. A signal may carry several different watermarks at the same time. Like traditional watermarks. Unlike metadata that is added to the carrier signal. The needed properties of a digital watermark depend on the use case in which it is applied. It is typically used to identify ownership of the copyright of such signal. 2. for example. With identical and highly similar blocks neighboring each other in the re-ordered sequence. but does not need to contain a relation to the carrier signal.e. i. a digital watermark has to be rather robust against modifications that can be applied to the carrier signal. if integrity has to be ensured. For marking media files with copyright information. video. whereas in digital watermarking. the signal may be audio. DIGITAL WATERMARKING A digital watermark is a kind of marker covertly embedded in a noise-tolerant signal such as audio or image data.

If a copy of the work is found later. and therefore removal. Digital watermarking is the process of inserting a digital signal or pattern (indicative of the owner of the content) into digital content. then the watermark may be retrieved from the copy and the source of the distribution is known. it should be robust. quality. These are known as perceptually based watermarking techniques. can be used later to identify the owner of the work.Both steganography and digital watermarking employ steganographic techniques to embed data covertly in noisy signals. the owner can still be identified. Watermarks of varying degrees of obtrusiveness are added to presentation media as a guarantee of authenticity. and common signal processing alterations used to make the data more efficient to store and transmit.1 Principle of digital watermarks A watermark on a bank note has a different transparency than the rest of the note when a light is shined on it. ownership. by pirates less possible. falling into 2 main categories. this method is useless in the digital world. including intentional attacks to remove the watermark. Since a digital copy of data is the same as the original. The effectiveness of a watermark is improved when the technique exploits known properties of the human visual system. Transparency requires a watermark to be imperceptible so that it does not affect the quality of the content. and to trace illegal copies of the work.1. digital watermarking is a passive protection tool. but does not degrade it nor controls access to the data. digital watermarking tries to control the robustness as top priority. In particular. One application of digital watermarking is source tracking. to authenticate the content. depending on in which domain the watermark is constructed: the spatial domain (producing spatial watermarks) and the frequency domain (producing spectral watermarks). and makes detection. a watermark should adhere to a few requirements. There are a variety of image watermarking techniques. A watermark is embedded into a digital signal at each point of distribution. .2. But whereas steganography aims for imperceptibility to human senses. This is so that afterwards. To be effective in its purpose. and transparent. Within this category. the class of image-adaptive watermarks proves most effective. known as a watermark. However. and source. The signal. The media of focus in this paper is the still image. . It just marks data. This technique reportedly has been used to detect the source of illegally copied movies. Robustness requires that it be able to survive any alterations or distortions that the watermarked content may undergo.

and is therefore probable that duplication on the Internet occurs without the rightful owners' permission. including Internet imaging. An example of an area where copyright protection needs to be enforced is in the on-line music industry. medical imaging. so that the source as well as the intended recipient of the digital work is known. Digital watermarking is the content protection method for the multimedia era. Digital watermarks are added to images or audio data in such a way that they are invisible or inaudible Ñ unidentifiable by human eye or ear. and that digital watermarking is needed to indicate who the culprit is. the images must be watermarked as they are captured. 2. This suggests that there are many applications that could require image watermarking. Copyright owners can incorporate .1.1. these same opportunities provide ease of access to these works. including writers. 2. which has resulted in pirating. RIAA reports that "record labels see watermarking as a crucial piece of the copy protection system. Embedding a watermark should not result in a significant increase or reduction in the original data.Currently there are various techniques for embedding digital watermarks. surveillance imaging. It is easy to duplicate audio and visual files. sooner or later. Furthermore. digital libraries. they all digitally write desired information directly onto images or audio data in such a manner that the images or audio data are not damaged. whether their music is released over the Internet or on DVD-Audio".2 IMPORTANCE OF DIGITAL WATERMARKS The Internet has provided worldwide publishing opportunities to creators of various works. Another scenario in which the enforcement of copyright is needed is in newsgathering. image and video databases. they can be embedded in content with a variety of file formats. They are of the opinion that any encryption system can be broken.3 PURPOSES OF DIGITAL WATERMARKS Watermarks are a way of dealing with the problems mentioned above by providing a number of services:  They aim to mark digital data permanently and unalterably. digital cameras. musicians and artists. Digital watermarking is being recognized as a way for improving this situation. This is so that later. video-ondemand systems. and satellite-delivered video. The Recording Industry Association of America (RIAA) says that the value of illegal copies of music that are distributed over the Internet could reach $2 billion a year. Basically. When digital cameras are used to snapshot an event. image's origin and content can be verified. photographers. However.

That is. watermarks are used in the protection of ownership. they demonstrate the quality and assure the authenticity of the work. Addition of a constant offset to the pixel values Local exchange of pixels other intentional attacks: Printing and Rescanning Watermarking of watermarked image (rewatermarking) 2. because each purchaser of the data has a unique watermark embedded in his/her copy.  Some more recent techniques are able to correct the alteration as well. Requantization.1. scaling and cropping. owners are able to find illegal copies of their work on the Internet. A/D conversion Resampling. Recompression Linear filtering such as high pass and low pass filtering. translation.5 DIGITAL WATERMARKING APPLICATIONS Digital watermarking is rapid evolving field. D/A conversion. Geometric Distortions: include such operations as rotation.  Watermarks can be used to identify any changes that have been made to the watermarked data. The presence of a watermark in a work suspected of having been copied can prove that it has been copied. this section identifies digital watermarking applications and provides an overview of digital watermarking capabilities and useful benefits to customers. 2.identifying information into their work.4 ATTACKS ON WATERMARKS            Lossy Compression: Many compression schemes like JPEG and MPEG can potentially degrade the data’s quality through irretrievable loss of data.  By indicating the owner of the work. Common Signal Processing Operations: They include the followings.1. any unauthorized copies that s/he has distributed can be traced back to him/her. In addition. The various applications are:    Authentication Broadcast Monitoring Copy Prevention .  With a tracking service.

audio. further.1. embedding and detecting watermarks.[2][3] In some countries. including the United States. embedded messages and more. so that even if the private key is exposed. Digital signature schemes in the sense used here are cryptographically based. the signature is valid. Digimarc: For document verification. Digital signatures are equivalent to traditional handwritten signatures in many respects.2 DIGITAL SIGNATURE A digital signature or digital signature scheme is a mathematical scheme for demonstrating the authenticity of a digital message or document. 2. A valid digital signature gives a recipient reason to believe that the message was created by a known sender such that they cannot deny sending it (authentication and non-repudiation) and that the message was not altered in transit (integrity). financial transactions. MediaSec: Provide software for various media types. a broader term that refers to any electronic data that carries the intent of a signature. Stegnosign: For creating. Digital signatures are commonly used for software distribution. Digital signatures employ a type of asymmetric cryptography. a properly implemented digital signature gives the receiver reason to believe the message was sent by the claimed sender. Digital signatures can also provide nonrepudiation.6 WATERMARKING SOFTWARE&SREVICES      Alpha-Tec: watermarking software for copyright protection and infringement tracking. For messages sent through a nonsecure channel.[4] and members of the European Union. and must be implemented properly to be effective. partial encryption.t. India. electronic signatures have legal significance. some non-repudiation schemes offer a time stamp for the digital signature. video e. copyright protection. and in other cases where it is important to detect forgery or tampering. and internet tracking. Digital signatures are often used to implement electronic signatures. but properly implemented digital signatures are more difficult to forge than the handwritten type. . Signum: Allow digital fingerprints to be embedded into grahics. while also claiming their private key remains secret.[1] but not all electronic signatures use digital signatures. meaning that the signer cannot successfully claim they did not sign a message.  Forensic Tracking E-Commerce/Linking 2.c.

identity. that information may not be accurate. When ownership of a digital signature secret key is bound to a specific user. a signature generated from a fixed message and fixed private key should verify the authenticity of that message by using the corresponding public key. Two main properties are required. The algorithm outputs the private key and a corresponding public key.1. digital signatures can provide added assurances of the evidence to provenance. Below are some common reasons for applying a digital signature to communications: 2. and status of an electronic document as well as acknowledging informed consent and approval by a signatory.  A signature verifying algorithm that. The United States Government Printing Office (GPO) publishes electronic versions of the budget. public key and a signature. given a message and a private key. A digital signature scheme typically consists of three algorithms:  A key generation algorithm that selects a private key uniformly at random from a set of possible private keys.  A signing algorithm that. University of Chicago. Secondly. 2. Digital signatures can be used to authenticate the source of messages. contracts. produces a signature. and Stanford are publishing electronic student transcripts with digital signatures. Universities including Penn State. or a message sent via some other cryptographic protocol. For example.2. either accepts or rejects the message's claim to authenticity.Digitally signed messages may be anything representable as a bitstring: examples include electronic mail. The importance of high confidence in sender authenticity is especially obvious in a financial context. suppose a bank's branch office . a valid signature shows that the message was sent by that user.1 Uses of Digital Signature As organizations move away from paper documents with ink signatures or authenticity stamps.1 Authentication Although messages may often include information about the entity sending a message. it should be computationally infeasible to generate a valid signature for a party who does not possess the private key. public and private laws. given a message. and congressional bills with digital signatures. First.2.

2. it may be possible to change an encrypted message without understanding it. if a message is digitally signed. known as nonmalleable ones.) However. The device signature may be in the form of  sensor pattern noise (SPN)  camera response function  Re sampling artifacts  Color filter array  Interpolation artifacts  JPEG compression  Lens aberration  sensor dust . because this is still considered to be computationally infeasible by most cryptographic hash functions (see collision resistance). If the central office is not convinced that such a message is truly sent from an authorized source.3 Non-repudiation Non-repudiation. any change in the message after signature will invalidate the signature. (Some encryption algorithms.2 Integrity In many scenarios. acting on such a request could be a grave mistake. Furthermore.1. but others do not.1. or more specifically non-repudiation of origin.sends instructions to the central office requesting a change in the balance of an account. Similarly.2. Although encryption hides the contents of a message. access to the public key only does not enable a fraudulent party to fake a valid signature. 2. the sender and receiver of a message may have a need for confidence that the message has not been altered during transmission. an entity that has signed some information cannot at a later time deny having signed it. By this property. there is no efficient way to modify a message and its signature to produce a new message with a valid signature. is an important aspect of digital signatures. prevent this. 2.

In photography. The raw image data captured by the image sensor is then converted to a full-color image (with intensities of all three primary colors represented at each pixel) by a demosaicing algorithm which is tailored for each type of color filter. and therefore cannot separate color information. so a color translation is required to convert the tristimulus values into a common. absolute color space. The color filters filter the light by wavelength range. is a mosaic of tiny color filters placed over the pixel sensors of an image sensor to capture color information. or color filter mosaic (CFM). Each two-by-two submosaic contains 2 green. The sensor's passbandquantum efficiency and span of the CFA's spectral responses are typically wider than the visible spectrum. Color filter array The Bayer color filter mosaic. such that the separate filtered intensities include information about the color of light. Since sensors are made of semiconductors they obey solid-state physics. For example. a color filter array (CFA). The responses of the filters do not generally correspond to the CIEcolor matching functions.CHAPTER 3 COLOR FILTER ARRAY 3. thus all visible colors can be distinguished. the Bayer filter (shown to the right) gives information about the intensity of light in red. green. and blue (RGB) wavelength regions. each covering one pixel sensor. 1 blue and 1 red filter. The spectral transmittance of the CFA elements along with the demosaicing algorithm jointly determine the color rendition. . Color filters are needed because the typical photosensors detect light intensity with little or no wavelength specificity.

Dick Merrill of Foveon distinguishes the approaches as "vertical color filter" for the Foveon X3 versus "lateral color filter" for the CFA. One cyan. and two green. used in a few cameras of Kodak.1 Manufacture of the CFA . (See Bayer filter#Alternatives) 4×4 2×4 3. green. and one magenta. with 50% white. With one blue. Bayer filter RGBE filter CYYM filter CYGM filter RGBW Bayer RGBW #1 2×2 2×2 2×2 2×2 RGBW #2 RGBW #3 Three example RGBW filters from Kodak. One cyan. used in a few Sony cameras. one yellow.  List of color filter arrays Pattern (pixels) 2×2 size Image Name Description Very common RGB filter. and red sensors on top of each other. used in a few cameras. and one magenta. one red. This arrangement does not require a demosaicing algorithm because each pixel has information about each color. one green. Bayer-like with one of the green filters modified to "emerald".The Foveon X3 sensor uses a different structure such that a pixel utilizes properties of multijunctions to stack blue. Traditional RGBW similar to Bayer and RGBE patterns. two yellow.

Aoki reveals that a CYWG arrangement was used. the individual CFA filters are usually layers of transmissive (absorptive) organic or pigment dyes. types. .". while other sensors have the CFA manufactured directly on the surface of the imager. Ensuring that the dyes have the right mechanical properties—such as ease of application. CMCR101B. In either case. Nakamura said that materials for on-chip color filter arrays fall into two categories: pigment and dye. attending optical properties. CMCR106R." He provides a bibliography focusing on the number. and subsequently dyed. Pigment based CFAs have become the dominant option because they offer higher heat resistance and light resistance compared to dye based CFAs. Some sources indicate that the CFA can be manufactured separately and affixed after the sensor has been manufactured. 1978) and not as a hybrid. Filter materials are manufacturer specific. For instance. At least one early example of an on-chip design utilized gelatin filters (Aoki et al. and spatial frequencies of the absorptive filters. Theuwissen says "Previously. thicknesses ranging up to 1 micrometre are readily available. and optimal manufacturing processes of color filter arrays. This makes it difficult. and resistance to humidity and other atmospheric stresses—is a challenging task. 1982).[15] The gelatin is sectionalized. at best. but nowadays. to fine-tune the spectral responsivities. First. via photolithography. and CMCR106B. aliasing effects. Theuwissen makes no mention of the materials utilized in CFA manufacture. moire patterns. There is some interference between the dyes and the ultraviolet light needed to properly expose the polymer. the color filter was fabricated on a separate glass plate and glued to the CCD (Ishikawa 1981). state "Several factors influence the CFA's design.Diazonaphthoquinone (DNQ)-novolacphotoresist is one material used as the carrier for making color filters from color dyes. CMCR106G. with the G filter being an overlap of the Y and C filters. durability. A few sources discuss other specific chemical substances.. though solutions have been found for this problem. CMCR101G. Adams et al. Color photoresists sometimes used include those with chemical monikers CMCR101R. all single-chip color cameras are provided with an imager which has a color filter on-chip processed (Dillon.

Solvent Yellow 88. 3. or Colour Index numbers. #12715. CAS Registry numbers.501. Image noise is an undesirable by-product of image capture that adds spurious and extraneous information.# 4. AKA Solvent Red 8. Ocean Optics has indicated that their patented dichroic filter CFA process (alternating thin films of ZnS and Cryolite) can be applied to spectroscopic CCDs. organics would be preferred over glass.2 Some pigment and dye molecules used in CFAs In U.3 IMAGE NOISE Image noise is random (not present in the object imaged) variation of brightness or color information in images.I. list some 150-200 chemical structures.Given that the CFAs are deposited on the image sensor surface at the BEOL (back end of line.096. # 61551.808.P. some CVD silicon oxide processes are low temperature processes. and is usually an aspect of electronic noise. It can be produced by the sensor and circuitry of a scanner or digital camera.. and C.S. In U.I.S. mainly azo dyes and pyrazolone-diazenyl. but fail to provide chemical names. where a low-temperature regime must be rigidly observed (due to the low melting temperature of the aluminum metalized "wires" and the substrate mobility of the dopants implanted within the bulk silicon). Solvent Blue 36. Carl Chiulli cites the use of 5 chemicals.801 Koya et al. # 5. the later stages of the integrated circuit manufacturing line). three of which are C.P. On the other hand. Gersteltec sells photoresists that possesses color filter properties. . of Fuji Photo Film company. 3. Image noise can also originate in film grain and in the unavoidable shot noise of an ideal photon detector.

including that which comes from the reset noise of capacitors ("kTC noise"). independent at each pixel and independent of the signal intensity." Image noise is. 3.4. caused primarily by Johnson–Nyquist noise (thermal noise). there can be more noise in the blue channel. inaudible. In color cameras where more amplification is used in the blue color channel than in the green or red channel. of course. Amplifier noise is a major part of the "read noise" of an image sensor. The magnitude of image noise can range from almost imperceptible specks on a digital photograph taken in good light. Gaussian. that is.4.2 Salt-and-pepper noise . of the constant noise level in dark areas of the image.4 Types o o o o o o o Amplifier noise (Gaussian noise) Salt-and-pepper noise Shot noise Dark current noise Quantization noise (uniform noise) Read noise Anisotropic noise 3. 3.Noise clearly visible in an image from a digital camera The original meaning of "noise" was and remains "unwanted signal". By analogy unwanted electrical fluctuations themselves came to be known as "noise.1 Amplifier noise (Gaussian noise) The standard model of amplifier noise is additive. from which a small amount of information can be derived by sophisticated processing (a noise level that would be totally unacceptable in a photograph since it would be impossible to determine even what the subject was). to optical and radioastronomical images that are almost entirely noise. unwanted electrical fluctuations in signals received by AM radios caused audible acoustic noise ("static").

that is. An image containing salt-and-pepper noise will have dark pixels in bright regions and bright pixels in dark regions. Dead pixels in an LCD monitor produce a similar. of the leakage. etc. The variable dark charge of normal and hot pixels can be subtracted off (using "dark frame subtraction"). and hot pixels appear as salt-and-pepper noise. variation in the number of photons sensed at a given exposure level. In addition to photon shot noise. there can be additional shot noise from the dark leakage current in the image sensor. This noise is known as photon shot noise. Shot noise has a root-mean-square value proportional to the square root of the image intensity. this noise is sometimes known as "dark shot noise" or "dark-current shot noise". but non-random. This type of noise can be caused by analog-to-digital converter errors. If dark-frame subtraction is not done. display. bit errors in transmission. or random component. 3.Image with salt and pepper noise Fat-tail distributed or "impulsive" noise is sometimes called salt-and-pepper noise or spike noise. . leaving only the shot noise. the noise will be more than just shot noise.3 Shot noise The dominant noise in the lighter parts of an image from an image sensor is typically that caused by statistical quantum fluctuations. and the noises at different pixels are independent of one another. It can be mostly eliminated by using dark frame subtraction and interpolating around dark/bright pixels. or if the exposure time is long enough that the hot pixel charge exceeds the linear charge capacity. Dark current is greatest at "hot pixels" within the image sensor. which is usually not very different from Gaussian.4. Shot noise follows a Poisson distribution.

robbing electrons of the thermal energy required to reach an intermediate state.4.[13] 3. with states near mid-band generating most of the dark current.6 Read noise Read noise is a property that is inherent to the CCD of digital cameras. For this reason. It has an approximately uniform distribution. The read noise of a camera affects how well the image represents the actual data. from which they are emitted into the conduction band. providing a path for valence electrons to sneak into the conduction band.4. and is present in all images taken and recorded by a camera. 3.4. Calibrating the read noise allows us know more about the quality of the CCD as well as the data distortion due to the reading of images.7 Anisotropic noise Some noise sources show up with a significant orientation in images. 3.3.4.4 Dark current noise: Dark current is the result of imperfections or impurities in the depleted bulk silicon or at the siliconsilicon dioxide interface. adding to the signal measured in the pixel. 3. or if dithering is explicitly applied. Though it can be signal dependent.5 In digital cameras . since high read noise decreases the quality of the image. the most effective way to reduce dark current is to cool the CCD. The generation of dark current is a thermal process wherein electrons use thermal energy to hop to an intermediate state.5 Quantization noise (uniform noise) The noise caused by quantizing the pixels of a sensed image to a number of discrete levels is known as quantization noise. image sensors are sometimes subject to row noise or column noise. These sites introduce electronic states in the forbidden gap which act as steps between the valence and conduction bands. it will be signal independent if other noise sources are big enough to cause dithering. For example. The efficiency of a generation center depends on its energy level.

The image on the right has adequate lighting and 0. That is.1 second exposure. this salt-and-pepper noise can be mostly eliminated by dark frame subtraction. can be introduced through brightening shadows or through color-balance processing. On most cameras. correct exposure requires the use of long shutter speeds. for a constant f-number. higher gain (ISO sensitivity). assuming the aperture area is proportional to sensor area. where read noise (noise floor) is significant. so larger sensors typically create lower noise images than smaller sensors. the sensitivity of an imager scales roughly with the sensor area. corresponding to increased ISO sensitivity. longer shutter speeds lead to increased salt-and-pepper noise due to photodiodeleakage currents.Image on the left has exposure time of >10 seconds in low light. In low light. In the case of images bright enough to be in the shot noise limited regime. The relative effect of both read noise and shot noise increase as the exposure is reduced. when the image is scaled to the same size on screen. . since fewer photons are counted (shot noise) and since more amplification of the signal is necessary. At the cost of a doubling of read noise variance (41% increase in read noise standard deviation). more pixels within a given sensor area will make the image noisier if the per pixel read noise is the same. not how this area is divided into pixels. For images at lower signal levels (higher ISO settings). or that the f-number or focal-plane illuminance is held constant. or printed at the same size. similar to shadow noise. Banding noise. is the largest determinant of signal levels that determine signal-to-noise ratio and hence apparent noise levels. the pixel count makes little difference to perceptible noise levels – the noise depends primarily on sensor area. 3. or both.6 Effects of sensor size The size of the image sensor. or effective light collection area per pixel sensor.

However.7 Sensor heat Temperature can also have an effect on the amount of noise produced by an image sensor due to leakage. But a definitive answer is not available. A simplified example of the impossibility of unambiguous noise reduction: an area of uniform red in an image might have a very small black part. If this is a single pixel. the noise level produced by a Four Thirds sensor at ISO 800 is roughly equivalent to that produced by a full frame sensor (with roughly four times the area) at ISO 3200. no algorithm can make this judgment perfectly. and that produced by a 1/2. This decision can be assisted by knowing the characteristics of the source image and of human vision.5" compact camera sensor (with roughly 1/16 the area) at ISO 100. since there is little important fine chroma detail that one risks losing. Furthermore. This ability to produce acceptable images at higher sensitivities is a major factor driving the adoption of DSLR cameras. so there is often a tradeoff made between noise removal and preservation of fine. it is likely (but not certain) to be spurious and noise. it is known that DSLRs will produce more noise during summer than winter. it may be a defect in a group of pixels in the image-taking sensor (spurious and unwanted. There are many procedures for this. many people find . An example shows a DSLR sensor at ISO 400 creating less noise than a point-and-shoot sensor at ISO 100. but not strictly noise). 3. With this in mind. which tend to use larger sensors than compacts.For instance. Most noise reduction algorithms perform much more aggressive chroma noise reduction. Many cameras have settings to control the aggressiveness of the in-camera noise reduction. involve some form of noise reduction. if it covers a few pixels in an absolutely regular shape. it may be more likely to be a true feature of the image.8 Image noise reduction Most algorithms for converting image sensor data to an image. 3. and average out the former while attempting to preserve the latter. low-contrast detail that may have characteristics similar to noise. but all attempt to determine whether the actual differences in pixel values constitute noise or real photographic detail. whether in-camera or on a computer. if it is irregular.

it is not so much a matter of noise reduction as of extracting a little information buried in a lot of noise. Since noise levels increase as ISO sensitivity is increased. such as astronomical images of very distant objects. The high sensitivity image quality of a given camera (or RAW development workflow) may depend greatly on the quality of the algorithm used for noise reduction. This leads to a breakdown of image quality at higher sensitivities in two ways: noise levels increase and fine detail is smoothed out by the more aggressive noise reduction. seeking small regularities in massively random data.luminance noise less objectionable to the eye. most camera manufacturers increase the noise reduction aggressiveness automatically at higher sensitivities. . techniques are different. In cases of extreme noise. since its textured appearance mimics the appearance of film grain.

occurring only at certain. etc. Although FPN does not change appreciably across a series of captures. . imager temperature. a long exposure (integration time) emphasizes the inherent differences in pixel response so they may become a visible defect. The term "fixed pattern noise" usually refers to two parameters. which describes the gain or ratio between optical power on a pixel versus the electrical signal output. which is the offset from the average across the imaging array at a particular setting (temperature. imager gain and incident illumination. pixel dependent photo response nonlinearity (PRNL) and is often simplified as a single value measured at almost saturation level to permit a linear approximation of the non-linear pixel response. The latter can be described as the local. degrading the image. integration time) but no external illumination and the PRNU (photo response nonuniformity). Pixel noise therefore gives a number (commonly expressed in rms) that identifies FPN in all permitted imaging conditions. it is not expressed in a random (uncorrelated or changing) spatial distribution.[1] One is the DSNU (dark signal non-uniformity).CHAPTER 4 FIXED PATTERN NOISE Fixed pattern noise is the term given to a particular noise pattern on digital imaging sensors often noticeable during longer exposure shots where particular pixels are susceptible to giving brighter intensities above the general background noise. It is characterized by the same pattern of 'hot' (brighter) and cold (darker) pixels occurring with images taken under the same illumination conditions in an imaging array. This problem arises from small differences in the individual responsibility of the sensor array (including any local post amplification stages) that might be caused by variations in the pixel size. material or interference with the local circuitry. Sometimes pixel noise[2] as the average deviation from the array average under different illumination and temperature conditions is specified. and fixed pixel locations. It might be affected by changes in the environment like different temperatures. it may vary with integration time. exposure times. In practice. which might strongly deteriorate if additional electrical gain (and noise) is included. Fixed pattern noise (FPN) is a general term that identifies a temporally constant lateral nonuniformity (forming a constant pattern) in an imaging system with multiple detector or picture elements (pixels).

1 PRNU (Photo Response Non-Uniformity) 4. However.1 Background Photo Response Non-Uniformity. produced by the Optoliner. The calculation of the PRNU is as follows:  Obtain the average image over the 100 images taken:   Subtract the DSNU image from this average image to eliminate the contribution from the DSNU. Obtain the spatial variance of the pixel values over the entire CCD   Divide the spatial variance by the average image from (ii) to obtain the PRNU as a percentage of the actual pixel values. We took 100 exposures each for three exposure times: 1/10.1. 1/4 and 1/2.00 candelas since the brighter light is more easily detected by the camera. PRNU is the pixel variation under illumination. In our measurements. is one source of pattern noise in digital cameras. 4. since increasing the illumination level will enhance the difference in the photo-response of the pixels across the image and lead to a higher PRNU. We expect the PRNU to increase with increasing illumination. we use the camera to take multiple images of a uniform scene.5.2 Methods To characterize the PRNU. Like DSNU. and increasing the illumination level increases the non-uniformity of the illumination .1. We kept the illumination level fixed at 3. it is seen as the variation in pixel responsively over the CCD. One of the few engineering definitions for PRNU or "photoresponsenonuniformity" is in the photonics dictionary. since the maximum value of the Opt linear device is around 4 candelas. Repeat the calculations for the different exposure times to compare the PRNU. and we also checked to ensure that the camera is focused before taking the pictures. And it is for CCD only. 4. while DSNU occurs as a variation in pixel responsively when the CCD is not illuminated. or PRNU for short.

a colour interpolation function generates the electronic signals of the other two colour components for every pixel according to the colour intensities of the neighboring pixels. g is the colour channel gain. . (1) and keeping the first two terms of the expansion where  is the denoised image and is the ensemble of the noises. shot noise. these signals are stored in the camera’s memory in a customized format. For most digital cameras. The PRNU pattern noise K can then be formulated as . and B). primarily the JPEG format. This colour filtering is determined by the CFA.. This colour interpolation process is commonly known as demosaicking. and is the input signal of the scene.e. the PRNU can be contaminated by various types of noise introduced at different stages of the image acquisition process. (1). In Eq. As formulated in [11]. R. the signal will inevitably be distorted when passing through each process and these distortions result in slight differences between the scene and the camera-captured image. (= 0.455) is the gamma correction factor. stand for dark current. read-out noise and quantization (lossy Compression) noise. G. respectively. but for every pixel only therays of one colour component is passed through the CFA and subsequently converted into electronic signals by the sensor. Finally. K is the zero-mean multiplicative factor responsible for the PRNU. and . After the conversion. after applying Taylor expansion to Eq. (1). during the image acquisition process. a camera output model can be expressed as where I is the output image.produced by the Opt linear. Since is the dominating term in Eq. the lenses let through the rays of the three colour components of the scene. A colour photo is represented in three colour components (i. gamma correction and image enhancement. In acquiring an image. However.s andr are random noise and is the fixed pattern noise (FPN) that is associated with every camera and can be removed by subtracting a dark frame from the image taken by the same camera. including . Figure 1 demonstrates the image acquisition process. we chose to increase the exposure times to mimic the effect of increasing illumination levels. The signals then undergo additional signal processing such as white balance. The dominating component of sensor pattern noise is photo response non-uniformity (PRNU).

has been reported as effective in producing good results. the discrete wavelet transform followed by a Wiener filtering operation). the noise residual patterns are extracted using Eq. only one of the three colours of each pixel is physically captured by the sensor while the other two are artificially interpolated by the demosaicking process. as suggested in [11]. is the gamma correction factor . where is the mean function.. thus identifying the source device that has taken the image under investigation.. is used to compare the noise Note in Eq. 1) First. The normalised cross-correlation is the noise residual extracted from . it is important that the PRNU extracted is as close to the genuine pattern noise due to the sensor as possible. for each imaging device d.2 Use of PRNU in Device Identification The basic idea of using the PRNU noise pattern in device identification can be described as follows. (5) and compared against the reference PRNU Kd of each device d available to the investigator in the hope that it will match one of the reference fingerprints.e. Although various denoising filters can be used.e. (6) is element-wise. Again the multiplication operation in Eq. (5) from a number of low-contrast images taken by device d and then the PRNU is estimated using the ML estimation procedure adopted by Chen et. we used . Given the PRNU-based approaches‟ potential in resolving device identification problem to the accuracy at individual device level. (5) is element-wise. (6). the wavelet-based denoising process (i. 4.is the noise residual obtained by applying a denoising filter on image I. Since for most cameras. where S is the number of images involved in the calculation. instead of using against the reference fingerprint . the noise residual WI of image I under investigation is extracted using Eq. 2) Secondly.is the s-th image taken by device d and Note the multiplication operation in Eq. this inevitably introduce noise with power stronger than that of the genuine . al. i..

(4)). as mentioned earlier that I is to apply the discrete wavelet transform followed by a Wiener filtering operation directly to the entire image I without differentiating physical components from artificial components and. Because the PRNU is formulated in Eq.3 DEMOSAICING A demosaicing (also de-mosaicing or demosaicking) algorithm is a digital image process used to reconstruct a full color image from the incomplete color samples output from an image sensor overlaid with a color filter array (CFA). However.. Eq.PRNU. Addressing this shortcoming is the motivation of this work. so demosaicing is part of the processing pipeline required to render these images into a viewable format. In Section IV. (2). The algorithm should have the following traits:  Avoidance of the introduction of false color artifacts. Many modern digital cameras can save images in a raw format allowing the user to demosaic it using software. The aim of a demosaicing algorithm is to reconstruct a full color image (i. 4. such as chromatic aliases. Section V concludes this work. (3) and (5) as a function of the noise residual W (i. in the rest of the work we will use the two terms. PRNU and noise residual. rather than using the camera's built-in firmware.e. In this work. as a result. a full set of color triples) from the spatially under sampled color channels output from the CFA. we will look at the impact of demosaicking on PRNU fidelity in Section II and propose an improved formula for extracting PRNU in Section III. Most modern digital cameras acquire images using a single image sensor overlaid with a CFA. interchangeably whenever there is no need to differentiate them. allowing the interpolation noise in the artificial components to contaminate the real PRNU in the physical components. zippering (abrupt unnatural changes of intensity over a number of neighboring pixels) and purple fringing  Maximum preservation of the image resolution .e. We can see from Eq. It is also known as CFA interpolation or color reconstruction. (3) and (4) that the accuracy of both PRNU K and noise residual W depends on the denoising operation applied to I in obtaining the most common method of obtaining . we present some experiments on device identification and image content integrity verification to validate the proposed PRNU extractionformula.

each coefficient of the wavelet transform used in the noise residual extraction process involves multiple pixels and thus both artificial and physical components. To extract the PRNU using Eq. and is called demosaicing. demosaicking has been rigorously investigated. in one noise residual extraction process. As a result the interpolation noise gets diffused from the artificial components into the physical ones. However. hence the accuracy of colour interpolation result is also content-dependent. The main problem inherent to Eq.. e. thus the interpolation noise is also more significant. a form of interpolation is needed to fill in the blanks. in inhomogeneous areas. The mathematics here is subject to individual implementation. which contains both artificial and physical components. the interpolation function can more accurately generate artificial components. the discrete wavelet transform followed by a Wiener filtering operation is applied. because of the low variation of the colour intensities of neighbouring pixels.g. the existing method for extracting PRNU as formulated in Eq. the colour variation between neighbouring pixels is greater. we call the colour components physically captured by the sensor as physical colours and the ones artificially interpolated by the demosaicking function as artificial colours. Conversely. (4) and (5) based on the definition of the output image model in Eq. Most demosaicking approaches group the missing colours before applying an interpolation function. When wavelet transform is applied during the noise residual extraction process the interpolation noise residing in the artificial components propagates into the physical components. (4) and (5). in a homogeneous area. For example. However. The grouping process is usually content-dependent. For example. edge-adaptive or non-adaptive. Therefore it is desirable to devise a noise residual extraction method that can . (1) does not take this into account. in the red colour component/plane of an image taken by a camera with a Bayer CFA. 4. Due to the fact that demosaicking is a key deterministic process that affects the quality of colour images taken by many digital devices. only one fourth of the pixels‟ red colour are physical and for each pixel with physical red colour all its 8 neighbours‟ red colours are artificial. This indicates that the PRNU in physical colour components is more reliable than that in the artificial components. Low computational complexity for fast processing or efficient incamera hardware implementation  Amenability to analysis for accurate noise reduction To reconstruct a full color image from the data collected by the color filtering array.4 DEMOSAICKING IMPACT ON PRNU FIDELITY In this work. (4) is that it involves the whole image plane.

prevent the artificial components from contaminating the reliable PRNU residing in the physical components with the interpolation noise. .

to make the proposed CD-PRNU versatile and applicable to cameras adopting different CFA patterns. Otherwise. According to Eq. Let be an interpolation matrix with 2N+1 × 2N+1 coefficients and be a X × Y-pixel input signal from the scene consisting of three colour components. we makes no assumption about the CFA pattern. (7) means that if the colour component c is the same as the colour that the CFA pattern F allows to pass. For each colour component of a pixel . without prior knowledge about the CFA. (1) proposed in can be reformulated as . a mathematical model for the CD-PRNUis derived and then an extraction algorithm is proposed to extract the noise residual that is to be used for estimating the final CD-PRNU. That is to say that for each pixel . The other two colour components are to be determined by the demosaicking process. G (green) and B (blue) before colour interpolation. i. (7) is artificially applied to calculate the colour. 5. except that it is a 2 × 2 square array. can be determined according to The first part of Eq.CHAPTER 5 CD-PRNU (Color Decoupled Photo Response Non-Uniformity) 5. (7). R (red). F. First. we will discuss the formulation and extraction of CD-PRNU. Although the 2×2 Bayer CFA is the most common CFA pattern.e .1 FORMULATION OF COLOUR DECOUPLED PRNU (CD-PRNU) In this section. only one of the three colour components takes a value physically captured by the sensor and this colour is determined by the colour configuration of the CFA pattern F. then no demosaicking is needed because c has been physically captured by the sensor. the second part of Eq. the image output model of Eq.2 Mathematical Model of CD-PRNU A generic demosaicking process is to convolve an interpolation matrix with an image block of the same size centred at the pixel where the artificial colour is to be calculated.

which is actually the interpolation noise. That is where is a low-passed filtered version of the artificial components and is the corresponding ―sensor pattern noise‖.Eq. We can also use the same ML estimate as in Eq. We will discuss how the physical and artificial is the interpolation noise extracted from colour components can be decoupled in simple manner without a priori knowledge about the CFA pattern in Section III.. (9) that the physical components and artificial components have similar mathematical expression. the PRNU remains unaffected by the interpolation noise.e. the PRNU is actually the interpolation noise P while. P can be extracted in the same way as the sensor pattern noise K is extracted (i. Hence if the physical and artificial colour components can be separated / decoupled. in the physicalcomponents. Eq. (9) suggests that in the artificial components. It can also be seen from Eq. (3)).3 CD-PRNU Extraction Algorithm . 5. (5) to extract the reference interpolation noise d from S low-variation images taken by d such that for a particular device where is the artificial colour components of the s-th low-contrast image taken by device d and .B.

By decoupling the physical and artificial colour components in this fashion before extracting the noise residual. For each colour channel. (4) is then used to obtain noise residual from each sub-images . Note that Algorithm 1 is for extracting the noise residual pattern W from an image I. Most CFA patterns are of 2 × 2 elements and are periodically mapped to the sensors. Unfortunately. But by decomposing into four sub-images. we first separate the three colour channels of a colour image I of pixels. we know that each of the four sub-images either contains only the physical colour or only the artificial colours. for each pixel of I. for each channel . such that vertical dimensions to get four sub-images. several methods have been proposed to estimate the CFA. we can extract the sensor pattern noise and interpolation noise. without knowing the CFA pattern used by the manufacturer. The framework of the colour decoupled noise residual extraction process is shown in Figure 2 and the procedures are listed in Algorithm 1. respectively. Eq. . Finally the CD-PRNU Wc of each colour channel c is formed by combining the four sub-noise residuals such that where. .According to Eq. We know that. we can prevent the artificial components from contaminating the physical components during the DWT process. so the second step is. However. exhaustive search is by no means acceptable. (10) and (11). Therefore. we perform a 2:1 down-sampling across both horizontal and . To estimate the CD-PRNU Pd of a particular device d and use it as the reference signature of d. we do not know (actually we do not have to know) which pixels carry the colour captured physically by the hardware and which are not. and mod is the modulo operation. from the physical and artificial components if the CFA is known. to extract the CD-PRNU. However. In this work. (11) is applied. only one of the three colour components is physical and the other two are artificial. these methods have to exhaust all of the possible CFA patterns in order to infer/estimate the „real‟/optimal CFA. manufacturers usually do not provide information about the CFA used by their cameras. . Eq.

.

) of each camera Ci is generated by taking the weighted average of the PRNUs extracted from 30 photos of blue sky according to Eq.4 Algorithm 1.5 EXPERIMENTAL RESULTS In this section. we carry out experiments on source camera identification and image content integrity verification to validate the feasibility of the proposed CD-PRNU in a comparative manner. 5.5. Cameras used in the experiments. The six cameras are listed in Table1. Source Camera Identification We have carried out source camera identification tests on 300 2048×1536-pixel photos of natural scenes taken by six cameras(C1 to C6). (11). we need clean PRNUs (which appear as high frequency bands of images) as device . Noise residual extraction algorithm Input: original image I Output: colour decoupled noise residual W Noise residual extraction algorithm 5. The reference PRNU (i.5.e.1. For device identification purpose. each responsible for 50. Table 1.

which has been proved to be a more stable detection statistics than normalised cross-correlation when applied to the scenarios in which the images of interest may have undergone geometrical manipulations. other images with low-variation scenes (i. Generally speaking. such as rotation or scaling. Table 2 lists the identification rates. Actually. Therefore. a camera is identified as the source camera. Our empirical experience suggests that an average of 20 blue sky images is accurate enough. its reference PRNU (or CD-PRNU) is most similar to the PRNU (or CD-PRNU). C4. C3. normalised cross-correlation formulated as in Eq. the normalised cross-correlation has to be greater than a specified threshold for a camera to be identified as the source camera. However.0000). so geometrical transformations will not be applied in order to prevent biased evaluation from happening. C5 and C6 perform significantly better when CD-PRNU is used in all cases. except for a few cases when images are of full size (1536 × 2048 pixels) and the identification rates are close or equal to 100% (1. C1. Source camera identification requires similarity comparisons among PRNUs (CD-PRNUs) and therefore the feasibility of the chosen similarity metrics is important. the key point is about demonstrating the different performance of the traditional PRNU and the proposed CD-PRNU. Another reason is that. the less data is available. if out of the six reference PRNUs (or CDPRNUs). therefore identification results become less reliable. WI.e. of the image I under investigation. (6) will be used to measure the similarity between PRNUs (CD-PRNUs). we also compare the performance of the proposed CD-PRNU against that of the traditional PRNU [11] when they are applied to blocks of 5 different sizes cropped from the centre of the fullsized PRNU (CD-PRNU). We suspect that the reason C2 does not perform as expected is because the CFA pattern is not a 2 × 2 square array as we have assumed. Because PRNU is often used in content integrity verification. thus giving better chance of extracting clean PRNU. Taking the average of the PRNUs from 30 blue sky images is to further reduce variation. in the following experiments. The reason blue-sky images are chosen in this work is because blue sky contains less scene details (high frequency signal). in this experiment. Individually speaking. The purpose of this experiment is to demonstrate the capability of the proposed CD-PRNU in dealing with the colour interpolation noise.. Fridrich suggested the use of the Peak to Correlation Energy (PCE) measure in [15]. because the smaller the images. .fingerprints for comparison against the PRNU extracted from individual images under investigation. where smaller image blocks have to be analysed. For C2. PRNU performs equally well as CD-PRNU when the image size is 192 × 256 pixels and slightly outperforms CD-PRNU when the block size is 48 × 64 pixels. scenes without significant details) can be used instead. Therefore. In practice.

a ROC curve of the performance of PRNU and CD-PRNU are demonstrated.when the statistics of the six cameras are pooled together. . This has been graphically presented in Figure 3(a). Figure 3. we can see that CD-PRNU still outperforms PRNU significantly. as listed in the Total column of Table 2. Performance comparison of source camera identification a) Overall identification rates when CD-PRNU and PRNU are used as fingerprint In Figure 3(b). We can see that the CD-PRNU outperforms the PRNU because at all fixed False Positive rate the CDPRNU‟s True Positive rate are always higher than that of the PRNU.

567 seconds to calculate the similarity between a pair of CD-PRNUs of the same size. it takes 0. Performance comparison of source camera identification b) Overall ROC curve when CDPRNU and PRNU are used as fingerprint For a system with a Pentium Core II 1. these two operations are trivial and only incur negligible increase of time complexity.526 seconds to compute the similarity between the PRNUs of two images of 2048 × 1536 pixels and 0.3G CPU and 3 GB RAM. .Figure 3. Although extracting CD-PRNU requires down-sampling and up-sampling. The amount of data processed during the extraction of PRNU and CD-PRNU is the same.

1 in Figure 4(a). . 5. we copied a 160 × 390-pixel area from Image I. and pasted it at approximately the same location in Image I.2 in Figure 4(b) to create the forged Image I.  In the first experiment. The images in Figure 4(a) and (b) are taken by Olympus C730. Source camera identification rates using traditional PRNU and proposed CD-PRNU.Table 2.2 Content Integrity Verification We also carried out the following three content integrity verification experiments on 640 × 480-pixel images.3 as shown in Figure 4(c).5.

3 in Figure 6(c). which covers the face of the person. which is taken by Olympus C730. pasted it at the area where the face of another person is in Image III.3  In the second experiment.2 (c) Forged Image II. we slid a 128 × 128-pixel window across the PRNU extracted from the image under investigation and another window of the same size across the reference PRNU of the cameras that have taken images I. we cropped an 80 × 100-pixel area from Image II.1 in Figure 6(a) taken by Canon Power Shot A400.2.2 in Figure 5(b) to create the forged Image II. The images in Figure 5(a) and (b) are also taken by the same camera. to create the forged Image III. (a) Original Image III.1 (b) Original Image I.2 in Figure 6(b). pasted it at the area where the face of another person is in Image II. which covers the face of the person.1 in Figure 5(a). the windows are . (a) Original Image II. (a) Original Image I. source image and forged images for the content verification experiments. Figure 5. The original image.3 To detect the manipulated areas.2 (c) Forged Image III. In Chen’s method [11].2.2 and III. Figure 6. II.2 (c) Forged Image I.Figure 4.3 in Figure 5(c). The original image. The original image. we cropped a 60 × 80-pixel area from Image III. source image and forged images for the content verification experiments.3  In the third experiment. source image and forged images for the content verification experiments.1 (b) Original Image II.1 (b) Original Image III.

0 to 3.moved a pixel at a time. where and are the mean and standard deviation of the correlations distribution. . the cross-correlation of the PRNU patterns inside the two windows at the same location was calculated according to Eq. the cross-follows the Generalized Gaussian (GG) distribution. TN = |{B | M(B) = 0 and Md(B) = 0}|. respectively. TN and FN are defined as TP = |{B | M(B) = 1 and Md(B) = 1}|. FP = |{B | M(B) = 0 and Md(B) = 1}| and FN = |{B | M(B) = 1 and Md(B) = 0}|. we use various thresholds defined as to analyze the performance of PRNU and CD-PRNU. Table 3. and T(t) is the threshold. the block in the centre of the window is deemed as manipulated. In the following experiments we will allow t to vary independently in the range from 0. As discussed in [11]. in our experiment. Table 3 lists the number of manipulated and non-manipulated blocks of 5 × 5 pixels in the forged images. this method is not accurate at the pixel level [11]. By varying the value of t. and lower FP and FN indicate better performance. If the cross-correlation is lower than a predetermined threshold t. (6).0 and use the four metrics. false positive (FP). therefore. Therefore. Number of manipulated and non-manipulated areas in each image (unit: block). Moreover. while higher TN and FN. FP. true positive (TP). which incurs a high computational load. Let B be an arbitrary block and M(B) and Md(B) be defined as TP. we can evaluate the integrity verification performance across a wide range of correlation thresholds T(t). we will obtain lower TP and FP. Higher TP and TN. true negative (TN) and false negative (FN) to measure the performance of integrity verifications based on PRNU and CD-PRNU. the sliding step/displacement is set to 5 pixels in order to reduce the computational load without sacrificing the accuracy of the integrity verification. To decide whether a block centered at the window superposed on the image has been manipulated or not. As t grows.

The ROC curves for the integrity verification experiments on image I. The final authentication result is a image with the dilated areas highlighted as the tampered areas. we use the raw data without any filtering to calculate the TP. Chen applies erosion and dilation operations with a square kernel in order to filter small areas identified as tampered with. FP and FN. giving rise to a higher FP. therefore the PRNU pattern noises in the two areas are almost the same.1 is at approximately the same location as the original area in image I.2. However. It is clear that the ROC curve of the PRNU-based scheme mostly overlaps with that of Random Guess. Thus. In order to simplify the comparison and to obtain a fair result. leading to higher TP.3 Figure 7 shows the performance of the PRNU and CD-PRNU in terms of TP. which means the authentication result is generally as unreliable as that of a random guess. As a result.3 is illustrated as Figure 8. We can see from Figure 7(a) and 7(b) that CD-PRNU generally achieves higher TP and TN while maintaining lower FP and FN. an algorithm with better performance will have a higher true positive rate (). a low threshold also results in the situation where more authentic blocks are mistakenly detected as manipulated. the performance of the filtering / dilation operation strongly depends on parameter setting and hence many experiments must be run to obtain the best parameters for filtering. TN. As a result. Therefore a ROCcurve of TP rate with respect to FP rate can be used to evaluate the overall performance of the PRNU and CD-PRNU.3 demonstrate that CD-PRNU-based method significantly outperforms the PRNU-based method when the tampered area is about one quarter of the sliding window. TN. ―the block dimensions impose a lower bound on the size of tampered regions that our algorithm can identify. the scheme cannot detect the manipulated area based on PRNU. 5. Let α be the number of manipulated blocks and β be the number of authentic blocks.3 across a range of correlation threshold T(t). the experiments on III.2.1 Experiment on Image I. A lower correlation (similarity) allows the algorithm to detect more manipulated blocks. which is marked along the horizontal axis of the ROC curve. FP and FN when authentication is carried out on image I. the ROC is formulated as At the same false positive rate .5. However. By . which is marked vertically. This is because the area we copied from the source image I.According to Chen‟s predication. we remove all simply connected tampered regions from Z that contain less than 64×64 pixels (one quarter of the number of pixels in the block)‖.

0 to 3.2.contrast. 5. with t varying from 0.3 .5. The ROC curve of Truth Positive Rate with respect to False Positive Rate of PRNU and CD-PRNU when authentication is performed on image I. which means that by using CD-PRNU manipulated blocks can be detected more reliably.0.3. Authentication results on image I. across a range of correlation threshold T(t).3: Integrity verification performance of the PRNU and CD-PRNU in terms of a) TP.2 Experiment on Image II. Figure 8. b) TN. Figure 7. the CD-PRNU-based scheme results in a curve much higher than the PRNU-based method.

0 to 3. . However. mixed performance in terms of TN and FP can be seen in Figure 9(b) and 9(c).When verifying the integrity of image II.3: Integrity verification performance of the PRNU and CD-PRNU in terms of a) TP. b) TN.0. Albeit their mixed performance in terms of TN and FP. c) FP and d) FN across a range of correlation threshold T(t). Authentication results on image II. Figure 10 also shows that the ROC curve of CD-PRNU is still slightly higher than that of PRNU. both PRNU and CD-PRNU can effectively detect the manipulated blocks as their ROC curves have suggested in Figure 10. again indicate its superiority to PRNU. with t varying from 0.3. indicating a slightly better performance of CD-PRNU. Figure 9. as shown in Figure 9(a) and 9(d). CD-PRNU‟s consistently higher TP and lower FN.

and therefore areas smaller than this should be filtered in order to remove the falsely identified noise. approximately one quarter of the window. which is only about one quarter of the sliding window (128 × 128 pixels). Since the tampered area is 60 × 80 pixels. as demonstrated in Figure 11(a). the manipulated blocks can be effectively detected by the CD-PRNU-based scheme because the areas in question are from two images taken by different cameras and thus contain different interpolation noise. the method based on PRNU can perform no better than a random guess. as can be seen in Figure 11(b) and 11(c). the CD-PRNU-based method can identify smaller areas.3 conforms to Chen’s observation.3.2. 5. By contrast.1. CD-PRNU‟s significantly better performance in terms of TP and lower FN can still be seen again in Figure 11(a) and 11(d).5. This poor performance is also reflected in the PRNU’s ROC curve in Figure 12 and is due to the fact that he manipulated area is too small (60 × 80 pixels). . although the performance of PRNU and CD-PRNU in terms of TN and FP are mixed.e.3. The ROC curve of Truth Positive Rate with respect to False Positive Rate of PRNU and CD-PRNU when authentication is performed on image II. the PRNU cannot correctly detect any manipulated blocks (i.3 Experiment on Image III.Figure 10.3 When authenticating III. The experiment result on III. As a result. When the threshold t is higher than 1. Chen predicated in that one quarter of the sliding window is the lower bound on the size of tampered regions that our algorithm can identify. respectively.

c) FP and d) FN across a range of correlation threshold T(t). . Authentication results on image III.0. with t varying from 0.0 to 3.3: Integrity verification performance of the PRNU and CD-PRNU in terms of a) TP.Figure 11. b) TN.

Sign up to vote on this title
UsefulNot useful

Master Your Semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master Your Semester with a Special Offer from Scribd & The New York Times

Cancel anytime.