Comparing The Difference in Image Distortion Between CCD and CMOS Camera Sensors

By Alan Herbert Student Number: 10022940 Module code: Digital Image production CE00012-4 Tutor: Anthony Gregorey word count: 2500


Comparing the Difference in Image Distortion Between CCD and CMOS Camera Sensors
Contents Title Page……………………………………………Page 1 Contents Page……………………………………..Page 2 Introduction………………………………………..Page 3 Aims And Objectives…………………..………..Page 4 Research……………………………………….…….Page 4 Method……………………………………………….Page 7 Results………………………………………………..Page 8 Conclusion……………………..……….….…….….Page 9 Evaluation…………………………….…….……….Page 9 Reference List……………………………………...Page 10 Bibliography ……………………………………….Page 11


Light is the most important feature of both film and photography and for many years people have been trying to capture and replicate images accurately. The study of light sensitivity and photographic emulsions in 1876 by Ferdinand Hunter and Vero Charles Driffield lead to the development of he first numerical measurement of film speed to be created. ISO, the international standard for rating film speed, is a combination of the American rating system (ASA) and the German institute for standardization (DIN) Before digital cameras were developed, celluloid film was the preferred choice of medium for capturing images. For black and white images the celluloid film was coated in two chemicals, one that contains silver halide salts that are photosensitive. The halides, also known as “grains”, are suspended in a gelatin. Once exposed to the light the grains of silver that have hit by photos have undergone a photochemical reaction and later when mixed with a developer solution, the exposed areas are stripped away from the film, thus creating a negative of the image captured. Film speed plays an important role in photography and in film for capturing more detail in darker areas of the image without increasing the aperture or decreasing the shutter speed. For ‘slower’ film such as ISO 100 the silver halides crystals are very small, and when developed they produce a fine grain that wouldn’t be as noticeable. However because the halide crystals were smaller they need to be exposed to the light for longer. For the faster film speeds the halide crystals were much larger so that they’d capture more light and faster, but would start to become more noticeable on the image produced. In digital photography there are still ISO ranges, but the grains that appear in the image, often referred to as ‘noise’, are different to the silver halide crystals as seen in film photography. In any electronic device there will always be a certain amount of noise produced when transmitting or receiving a signal. SNR, signal to noise ratio, is the universal way of comparing the definition of the signal against the background noise. Noise produced in digital photography and film is produced when amplifying the sensitivity of the sensor, as the signal becomes more pronounced so does the noise when increasing the camera’s ISO.


Aims and Objectives
What I am aiming to achieve in this study is to comparing the difference in image distortion between different types of camera sensors. I want to see what areas of an image are most affected by noise and the different types of noise that can occur. My objectives are to compare different types of camera sensors mainly between the APS-C CMOS sensor and a sensor used in film cameras, which tend to use CCD’s.

The last few years’ digital technology has become cheaper, one of the reasons for falling prices in digital photography and film has been the introduction of CMOS sensor in to the majority of DSLR cameras.

CCD sensors
If you imagine a 2D array of millions of solar cells, each on takes a small portion of light from an image and transforms it in to a voltage. The way the voltages are read is where the sensors start to differ. In a CCD (charged couple deice) sensor, the value for each line of cells is transported across the chip and read in one corner of the array. An analogue to digital converter turns each pixel values into a digital value. [Referance1: Shrinil Sonl,, April 2011] The reason why CCD sensors are more expensive to manufacture is because that a special manufacturing process is require for when the value of each pixel is transported across the sensor to be read it does cause any interference to the other cells as they are recording new values. This lead to high quality images with very low noise distortions but the CCD sensors do use up around about 100 times more power than an equivalent CMOS sensor. [Referance1: Shrinil Sonl,, April 2011]

CMOS Sensors
A CMOS (complimentary metal-oxide semiconductor) sensor is built upon 300-mm wafers using 40-nm line geometries. and processes that allows each pixel to read individually. A statement by Ray Fontaine, a process analyst at Canadian firm Chip works Inc., from Electronic Design Journal issue 9/22/11, said that, “while CMOS sensors traditionally require four transistors per pixel, newer designs have made substantial reductions. In fact, Sony now employs a transistor-sharing scheme to bring the total down to 1.375 transistors per pixel.” [Reference 2: Rodger Allan , Electronic design Journal, Issue: 9/22/2011] On the other hand advancement in CMOS sensors happened recently from Germany’s Franhofer Institute for microelectronic Circuits and Systems (IMS). Using 0.5 m pixels, researchers developed a CMOS image sensor that operates at -40°C to 115°C, suiting it to harsh industrial and environmental conditions.


Where as the upper limit of the CCD image sensor is at around 60°C. however a quote from Werner Brockherde from the Electronic Design Journal issue 9/22/2011 in the Engineering Feature, he says that “The main issue with making the cameras that can operate at these high temperatures is the increase in dark current. ” [Reference 2: Rodger Allan, Electronic design Journal – Issue: 9/22/2011] A rise in just 8°C doubles the dark current which manifests itself in electrical noise, reducing the cameras dynamic range and affecting its ability to record images at low light levels. In addition, ghosting in the form of artifacts or fuzziness degrades image quality. [[Reference 2: Rodger Allan, Electronic design Journal – Issue: 9/22/2011]

Types of Image Noise
There are 3 main types of image distortion; random, banding and fixed pattern. Random Image Distortion Random image noise is one of the hardest to remove from an image, as it would be hard to predict where it may occur. It appears as colour discrepancies around where there is a change of intensity in the image. It is generally affected by a combination of exposure length and ISO speed. Banding Image Distortion Banding image distortion is dependent on what camera you have, as not all cameras will create it. It happens during the digital processing of the image as the camera takes the data created by the sensor. It becomes particularly noticeable when using high ISO speeds, adjusting the brightness of an image is and mostly appears in the shadows of an image. Fixed Image Distortion Fixed image distortion surround hot pixels. Hot pixels are more intense than other pixels surrounding it and are much brighter than the noise variances. Long exposures and high temperatures can cause fix pattern noise to appear. If pictures are taken under the same settings again and again then the hot pixels shall appear in the same places, however fixed pattern noise is the easiest to fix and remove. [Reference 3:, 2010]


Colour Processing
One possible way to record colour images using CCD sensors is to capture the colour on 3 different sensors; red, blue and green. The light is divided up through a prism so each channel is recorded separately, as seen in [FIG.1] [Reference 4: Ted Ellis,, 12/5/2006]

Another way using the CCD sensors is to use a rotating disc that has the 3 different filters on it and the image is registered on one sensor. The most common way that is used by both CCD and CMOS sensors is to apply filters to each pixel however there is again two different ways of doings this. One is filtering the colour as RGB (Red, Green, Blue) and the other is to filter using CMYG (Cyan, Magenta, Yellow, Green). The CMYG filter allows a “higher pixel signal due to its broader spectral band pass.” However filter does take additional processing which can noise, as the final image has to be converted to RGB. [Reference 6: “CCD and CMOS technology: Technical white paper “research paper written by Axis Communications in 2010] The Most common way of filtering the colours is the Bayer Array. This is where rows of alternating red-green and green-blue filters cover the photodiodes. Since the human eye is more sensitive to green than to the other two colours the Bayer system compensates this by doubling the amount of green filter in the arrays. [Reference 6: “CCD and CMOS technology: Technical white paper “research paper written by Axis Communications in 2010] FIG.2


The cameras that I shall be using in my experiments are:  Canon 60D, which uses a 18MP APS-CMOS sensor  Canon XL2, which uses 3-CCD system but images are recorded on to mini-DV tape. I want to see which camera records the image that can maintain a smooth high quality image. I want to try and test the cameras against 2 different types of noise distortion that may occur and see which one still maintains its image quality. For each of the experiments I shall take exactly the same image for both of cameras so I can compare how well each does in representing the shadows and highlights. Experiment 1: Random Image Distortion This type of image distortion occurs through a combination of a long exposure and a high ISO. For this experiment I set up the camera tripod and record a 5 second shot of a dark location that has a bright light source in the center. I’ll set both of the cameras to same or at least very similar setting to each other so that I have a fair comparison. Experiment 2: Banding Image Distortion Banding image distortion does occur in every camera, but again I’ll set the cameras to the same setting and get them both to record an image for the same amount of time. The cameras will be set to a long exposure (slow shutter speeds) and a high ISO. Afterwards in Photoshop I can boost the contrast of the images and look in the shads for the amount of gain in an area.


Below are the results from the test images. Because the cameras didn’t have the same measurement for the ISO or gain I change the experiment slightly so that it a comparison between the performance of the cameras when they’re set to their highest amount of light sensitivity.

Canon 60D

Canon XL2

The canon 60D manages to hold a clear image where the light doesn’t bleed out into other colours. With an increased ISO it’s harder to get a perfectly sharp image.

The Canon XL2 has a large halo of noise distortion around the light and the noise make its incredibly hard to have a sharp clear image.

The Canon 60D has a very small amount of noise in the black regions cause by the halo of light around the lamp. It’s also managed to keep the outline of the lamp very clear and sharp. The 60D does appear to suffer from banding image distortion, only a bit of random noise due to the high ISO and long exposure. 8

The Canon XL2 has suffers badly with a high amount of noise distortion. The sensors sensitivity has cause a halo effect around the lamp, which is mostly noise but doesn’t blend smoothly into the darker regions. The lamp also has some color bleeding from the intensity of the light, and doesn’t hold a sharp outline. The XL2 suffers from slight banding image distortion and a high amount of random image noise

After comparing the camera sensors the canon 60D that uses the CMOS sensor produces a much finer quality of image compared to the XL2 that uses a 3 CCD system. However I don’t feel that it was a very fair experiment as the 60D has a much new CMOS sensor that has more megapixels so the definition of the image is a lot better. Also they are two very different cameras, it’s hard to compare them exactly as they both have different mechanics in how they control the exposure of the image. The 60D is more photography base so has a great range of capabilities to control the exposure of the image; also the film speeds were measured in the ISO system unlike the XL2 where its gain was measured in ‘db’.

Evaluation and Recommendations
Even though The Canon XL2 has a 3 CCD sensor system I think that one of its biggest flaws is the medium which its writes too, mini DV tape. Mini DV tape is a magnetic tape that records the information of the image. However the image quality must be limited by the amount of data it can write to the tape per second, as there is a limited speed at which the tape can move. CCD sensors do produce a very fine image but I would have like to have tested equipment that recorded on to a digital medium like the 60D as it would make the comparison more equal, but I was limited by the equipment that I could use. One of the problems I encountered was that because the canon XL2 was the only camera that I could get hold of that had a CDD sensor the only way I could compare the images it recorded was to record a 5 second film clip, export it into final cut and then get a still from that to open in Photoshop. I feel that having an image run through so many different processes before I have chance to compare it properly to the Canon 60D will have ruined the integrity of its image quality. At the moment there is a lot of research and development happening with the CMOS sensors as people are constantly try to reduce the size of the transistors on the photodiodes as producing these types of sensors are cheaper which then is in the price range of most amateur and semi professional camera enthusiast which is a massively expanding market. Even though currently CCD sensors are probably the best quality sensors there is on the market, it won’t be long until CMOS sensors catch up. On the 24th August 2010, Canon Inc. announced the successful development of their new APS-H CMOS sensor, which has approximately 120-million pixel resolution. [Reference 5:, written on 24th August 2010] But these sensors are not yet in mass production for the public.


Reference lists for text
[Reference1]: tal/question362.htm web page – accessed on 9/11/11 – camera sensors [Reference 2]: Electronic design Journal – Issue: 9/22/2011 – page 26 – Engineering feature by Rodger Allan [Reference 3]: webpage - acessed on 10/11/11 – types of image distortion [Reference 4]: Image Recording - web page – accessed on 1/12/11 – comparing CCD’s and CMOS sensors [Reference 5]: - web page, written on 24th August 2010 – accessed on 11/12/11 –CMOS sensors [Reference 6]: “CCD and CMOS technology: Technical white paper “research paper written by Axis Communications in 2010

Bibliography - web page –acessed on 8/11/11 – photochemistry - Web page - accessed on 8/11/11 – image noise web page - accessed on 8/11/11 – sensor crop factor webpage – accessed on 9/11/11 – film speed development webpage – accessed on 9/11/11 – film speed development webpage – accessed on 9/11/11 – photochemistry

10 web page – accessed on 9/11/11 – image noise distortion Referance [1] 2.htm web page – accessed on 9/11/11 – camera sensors web page – accessed on 9/11/11 – sensor sizes Reference[4]: webpage - accessed on 10/11/11 – types of image distortion Reference [2]: “CCD and CMOS technology: Technical white paper “research paper written by Axis Communications in 2010 webpage – accessed on 8/12/11 – by Steven W. Smith PhD Digital Photography Answers! Certified Tech Tupport - book published 1999 – written by Dave Johnson – drawings [Reference 4]: Image Recording - web page – accessed on 1/12/11 – comparing CCD’s and CMOS sensors [Reference 6]: “CCD and CMOS technology: Technical white paper “research paper written by Axis Communications in 2010 Reference lists for images [FIG.1] Digital Photography Answers! Certified Tech Support - book published 1999 –written by Dave Johnson – drawings on CCD sensors [FIG.2[ “CCD and CMOS technology: Technical white paper “research paper written by Axis Communications in 2010