The admissibility of digital photographs in criminal cases.

Link/Page Citation
Representing today's trend in photography, digital cameras continue to rapidly r
eplace traditional film-based models. As prices keep dropping, consumer ownershi
p will become even more prevalent. Similarly, law enforcement agencies have begu
n to favor digital cameras--just the latest in a long line of technological inno
vations used by departments to collect and document evidence. Digital photograph
y offers law enforcement numerous benefits, including instant access to images,
rapid transportability of pictures within a department or to outside agencies, a
nd decreased cost and time as these cameras require no film development.
Of course, photographs--which generally hold substantial weight--serve as one of
the most effective forms of evidence in court. However, many individuals in the
legal community fear the potential abuse and manipulation of digital images. Th
erefore, they consider these pictures inadmissible under current evidentiary rul
es.
[ILLUSTRATION OMITTED]
To this end, an examination of the admissibility of film-based photographs and a
n analysis of cases, legislation, and legal commentary pertaining to digital pic
tures can provide valuable insight. Further, agencies can follow recommendations
as to how they can help ensure the admissibility of their digital photographs u
nder the law as it develops in the United States.
FILM-BASED PHOTOGRAPHS
People can manipulate film-based pictures. Throughout the photographic process,
an individual skilled in photography can alter the image. (1) For instance, whil
e taking a picture, a person can use a narrow f-stop and a fast shutter speed to
make a photograph taken during the day appear as if someone took it at night.

Individuals also can alter a photograph during the development stage. "Through t
he judicious selection of exposure times for the paper emulsions and filters to
screen selective color wavelengths, a skilled photographer can produce a differe
nt image from the one ... viewed through the camera's eyepiece. This image could
appear to the untrained eye to be ... perfectly legitimate ... yet in subtle wa
ys could be misleading in the jurisprudential context." (2) Also, during develop
ment, a technique known as crop and splice can change the picture. Using this me
thod, a person combines two negatives by cropping out a portion of one and splic
ing in its place part of another. (3)
Modifications of film-based photographs have presented problems for years. (4) "
The forensic ramifications ... are obvious. A skilled photographer could artfull
y assemble through either pre- or postphotographic processing a photograph that
could either be highly incriminating or exculpatory. Litigants could then offer
that photograph as evidence in support of their cause. Under the Rules of Eviden
ce, to authenticate a photograph, a witness need only to say that the photograph
accurately depicts the scene, object, or person. In this scenario, if a witness
were willing to deceive the court with a manipulated photograph, discovery of s
uch perjurious intent would be problematic." (5) However, someone suspecting man
ipulation of a picture always can ask for the negative to trace its origin.
Any party seeking to introduce a film-based photograph into evidence must demons
trate its relevancy (i.e., add to the likelihood that an event did or did not oc
cur) and authenticity (i.e., a knowledgeable person must verify the image's accu
racy). (6) For example, a detective photographs a drug deal. The picture depicts
two individuals exchanging a package. The prosecutor wants to enter the photo i
nto evidence at the criminal trial of the individual who received the drugs. The
picture is relevant because it shows the person present at the scene where the
deal occurred and in receipt of the package. To authenticate the photograph, the
prosecutor can place on the stand the detective who took the picture or any off
icer who witnessed the transaction and elicit that the image actually represents
the person, package, place, and time. After establishing the photograph's relev
ancy and authenticity, the prosecutor can move to admit it into evidence.

One additional rule exists that pertains to the admissibility of all photographs
. Under the Best Evidence Rule, to prove the content of a picture, courts genera
lly require the original--defined as the negative or any print therefrom. (7) Th
erein lies a major perceived problem with digital images: the absence of a trace
able origin to rely upon (i.e., no negative). (8)
DIGITAL PHOTOGRAPHS
Digital photographs include pictures processed by computer. They consist of pict
ure elements, or pixels--computer codes consisting of bits of information repres
enting specific colors, intensities, and locations. More pixels result in a shar
per and clearer image.
A digital camera works similarly to a film-based model. However, instead of usin
g light-sensitive film, it employs a light-sensitive chip--a charged coupled dev
ice (CCD). The CCD records the picture electronically as its light sensors captu
re, convert, and store the image as blue, green, and red pixels, which then beco
me saved in the camera as a computer-readable data file. Using specialized softw
are, a computer can reconstruct the image and display it on a monitor and route
it to a printer.
Concerning admissibility, people mainly fear that digital photographs can become
altered more easily than film-based images "to fabricate evidence for improper
purposes." (9) Some in the legal community feel that such dangers in digital pho
tography overall necessitate different treatment under the Rules of Evidence.
Certainly, software used to create digital photographs allows alteration of the
picture. "At worst, objects ... not in the original image can be added and those
that were there can be removed." (10) However, detection of a manipulated digit
al picture does not prove difficult. "Factors such as the density of the image (
based on light exposure), the shadows in the picture, existence or nonexistence
of splice lines, and continuity of the image" can be scrutinized. (11)
Compression represents a secondary concern pertaining to digital photographs. Wh
ile the amount of film limits the quantity of pictures taken with a traditional
model, "digital cameras allow users to choose the number of images they want to
capture and store." (12) The compression of data files allows digital camera use
rs to save more pictures, resulting in lower-quality photos because when "the us
er wants to view the image, the decompression process 'guesses' what information
was discarded to produce a complete image." (13)

ADMISSIBILITY OF DIGITAL IMAGES
Cases
Few cases directly address the admissibility of digital photographs in courts of
law. In fact, the author found only one court in the U.S. federal and state sys
tems that tackled the issue head-on. The many courts that have yet to address th
e subject largely must extrapolate from opinions pertaining to other issues conc
erning digital imaging.
[ILLUSTRATION OMITTED]
The Georgia Supreme Court case of Almond v. State dealt directly with the admiss
ibility of digital photographs. (14) In that case, a jury found Mastro Almond gu
ilty of malice murder and the sale of cocaine. On appeal, Almond raised the issu
e of digital images as evidence at his trial. The court stated that because the
record showed "that the pictures were introduced only after the prosecution prop
erly authenticated them as fair and truthful representations of what they purpor
ted to depict," they were properly admissible. (15) The Georgia Supreme Court di
d not provide any other guidelines for determining the admissibility of digital
photographs. In fact, the court went on to say that "[W]e are aware of no author
ity, and appellant cites none, for the proposition that the procedure for admitt
ing pictures should be any different when they were taken by a digital camera."
(16)
Although no other court has dealt directly with the admissibility of digital pho
tographs, opinions exist that can offer insight as to where many will stand on t
he issue. For example, in People v. Rodriquez, the New York Supreme Court, Appel
late Division, stated that the trial "court properly exercised its discretion in
admitting bank surveillance videotapes, and photographs made from those tapes,
without expert testimony about the digitizing process used at the FBI laboratory
to slow the tapes down and make still photos from them, since a bank employee r
esponsible for making the original tapes at the bank testified that he compared
the original and slowed-down tapes and that what was represented therein was ide
ntical except for speed." (17) The People v. Rodriquez holding indicates that th
e court seeks to ensure that an individual with first-hand knowledge of the phot
ographed scene attests to the picture's accuracy. Again, this demonstrates that
for admissibility, photographs must be relevant and authenticated.
The Washington Court of Appeals case of State v. Hayden represents an additional
example that provides insight into how another state may rule on the admissibil
ity of digital photographs. (18) The case mainly focused on the admissibility of
digital imaging used to enhance latent fingerprints and palm prints. The court
held that "[b]ecause there does not appear to be a significant dispute among qua
lified experts as to the validity of enhanced digital imaging performed by quali
fied experts using appropriate software, we conclude that the process is general
ly accepted in the relevant scientific community." (19)
Although State v. Hayden dealt with the admissibility of digital enhancement tec
hnology under the Frye Test--used to determine the admissibility of novel scient
ific evidence--the court made four important points that support the admissibili
ty of digital photographs in general: 1) digital photography is not a novel proc
ess; (20) 2) the high cost may have contributed to the delay of digital image en
hancement in forensic science; (21) 3) the court opined that digital photographs
have an advantage over analog film photographs because they "can capture approx
imately 16 million different colors and can differentiate between 256 shades of
gray"; (22) and 4) like film photographs, digital images work with light sensiti
vity, except that the "computer uses a chip and a hard drive in place of the cam
era's film." (23) Based upon the dicta provided, Washington courts seemingly wou
ld rule on the side of admissibility concerning digital photographs.
Recently, the Court of Appeals in California addressed the use of digital imagin
g to enhance a shoe print in a criminal case. (24) In People v. Perez, the court
of appeals accepted the trial court's statement that a particular brand of soft
ware "is not a scientific technique" but represents "just an easier way of devel
oping film, developing a picture. And it does it by means of digital imaging of
pixels. Digital imaging ... is accepted scientifically and has been for decades.
" (25) After reading People v. Perez and in light of the previous cases mentione
d, courts in California seemingly would consider digital photographs admissible.

Legislation
Alternatively, a legislator sponsored Wisconsin Assembly Bill 584, which "prohib
its the introduction of a photograph ... of a person, place, document ... or eve
nt to prove the content ... if that photograph ... is created or stored by data
in the form of numerical digits." (26) The legislator apparently "became upset w
hen high school students manipulated a digital photograph by putting heads on bo
dies of the opposite sex." (27) If this bill becomes law, digital photographs wi
ll not be admissible in Wisconsin courts.
Legislators in Hawaii also have concern about the admissibility of digital photo
graphs. However, rather than taking the extreme position of seeking a ban in cou
rts, the legislature directed the Hawaii Supreme Court to establish written proc
edures governing police use of digital photography in traffic accident reconstru
ctions. The directions to the Hawaii Supreme Court are contained in Hawaii House
Bill 1309, which states, "[a]lthough current rules do not preclude the admissio
n of digital photographs as evidentiary material, such admissibility is continge
nt upon the basic data and collection technique meeting a threshold requirement
of reliability that has not yet been established by the Hawaii Supreme Court's S
tanding Committee on the Rules of Evidence." (28)
Legal Commentary
Many individuals in the legal community remain largely unreceptive to allowing t
he admission of digital photographs under the current rules of evidence. One aut
hor stated that "[a]lthough photographs may be manipulated, the potential for ma
king subtle but significant alterations to digital images gives cause for concer
n that digital images may be unfit for use as evidence in a court of law" (29) a
nd proposed amending the current evidentiary system specifically to deal with di
gital imaging.
In another article voicing concern over the admissibility of digital photographs
under current evidentiary systems, the authors stated, "As noted, current princ
iples of authentication have developed partly in response to certain assumptions
about the inherent limitations of traditional media technologies. The degree to
which these assumptions are appropriate in the context of today's highly sophis
ticated multimedia tools is an open question posing challenges for advocates, ju
dges, experts, and legislative bodies alike." (30) And, another author noted tha
t "[w]hile advances in technology are generally viewed as positive within societ
y as a whole, the potential for incredible abuse associated with electronic phot
ography is, or should be, troubling to the legal profession in particular." (31)

A final author nicely summed up such concerns among those in the legal community
by saying, "As the conventional photograph goes the way of the horse-drawn carr
iage and the vinyl phonograph record, courts and legislatures will have to estab
lish procedures to assure the accuracy and integrity of visual evidence admitted
into legal proceedings. If existing doctrines cannot rise to the task, new doct
rines will have to develop." (32)
[ILLUSTRATION OMITTED]
As evidenced by these statements, not everyone in the legal community agrees wit
h any court decision admitting digital photographs under the current Rules of Ev
idence. At its October 18, 2002, meeting in Seattle, Washington, the Advisory Co
mmittee on the Federal Rules of Evidence considered the concerns of commentators
who argue that digital photographs should not be admitted under current evident
iary rules. The committee held a preliminary discussion on whether to amend Rule
901, the authentication requirement, or if a new rule proved necessary to deal
with digital photographs. Ultimately, the committee members were skeptical of th
e necessity of a new rule and felt that Rule 901 "was flexible enough to allow t
he judge to exercise discretion to assure that digital photographs are authentic
and have not been altered." (33) However, the committee did direct its reporter
to "prepare a background memorandum on the use of digital photographs as eviden
ce" so that it could consider changes to the rules in the fu-ture due to its "in
terest in assur-ing that the rules are updated when necessary to accommodate tec
hnological changes." (34)
As for the requirements of the Best Evidence Rule, a logical reading indicates t
hat digital photographs are admissible under that rule. Generally, it requires t
he original to prove the content of a writing, recording, or photograph. (35) Un
der the Best Evidence Rule, "[i]f data are stored in a computer or similar devic
e, any printout or other output readable by sight, shown to reflect the data acc
urately, is an 'original.'" (36) Therefore, a digital image downloaded to a comp
uter and subsequently printed would seem to qualify. (37)
RECOMMENDATIONS
At a minimum, agencies should establish standard operating procedures that focus
on two goals that will ensure the admissibility of their digital photographs in
court: 1) preserve the original and 2) follow a reliable process demonstrating
the integrity of the image. Ideally, departments will concentrate on "chain of c
ustody, image security, image enhancement, and release and availability of digit
al images." (38)
When attempting to preserve the original, unmanipulated image, agencies should s
tore it on a compact disc that can be written to only once and then is only read
able (i.e., a CDR, rather than a CD-RW). This ensures that no one can remove or
alter the data without copying the original. After capturing an image, agencies
should immediately transfer it to a CD-R and label the disc with the date, time,
and place the picture was taken; the individual who captured the image; and oth
er important information associated with the photo.
Additionally, agencies should preserve the digital image in its original file fo
rmat, (39) rather than compressing it for storage. This allows the camera to cap
ture and store the most information possible. When departments must enhance a pi
cture, they should create a new image file, saving it separately and not writing
over the original.
When establishing reliable procedures that demonstrate the integrity of images f
rom creation to admission into evidence, agencies must limit access to the files
. As one commentator stated, "[i]mage handling procedures should be standardized
and access to digital images should be strictly controlled." (40) The process u
sed "should be able to demonstrate: who took the picture and when, where and how
the image was stored, who had access to the image from the time it was taken th
rough the time it is introduced in court, and any details on whether or not the
image has been altered and how." (41)
In this regard, reliable procedures will help prevent challenges to admissibilit
y by defense counsel. Also, they will allow agencies to track who had access to
the photographs and what, if anything, was done with them. Of course, any reliab
le procedures must begin with preserving the original.
Also worthy of note, some law enforcement agencies use commercial photo labs for
developing and processing film. Following such a procedure opens up possible ch
allenges when departments seek to admit these pictures in court. (42) In this re
gard, digital images prove superior to film-based photographs because no one out
side the department handles them.
When digital imaging is considered for law enforcement, the concern of the admis
sibility of digital photographic evidence in court is often raised. The fact tha
t digital photographs are more easily altered than film-based photographs is usu
ally cited. Some even believe digital photographs are not admissible in court.
This article is presented in the hope of clearing up some of the confusion and m
isinformation about this issue. We will begin with the rules of evidence regardi
ng digital evidence.
The Federal Level
Federal Rules of Evidence, Article X (Contents of Writings, Recordings and Photo
graphs), Rule 101(1) defines writings and recordings to include magnetic, mechan
ical or electronic recordings. Rule 101(3) states that if data are stored in a c
omputer or similar device, any printout or other output readable by sight, shown
to reflect the data accurately, is an "original". Rule 101(4) states that a dup
licate is a counterpart produced by the same impression as the originalby mechani
cal or electronic re-recording, or by other equivalent techniques which accurate
ly reproduces the original. And Rule 103 (Admissibility of Duplicates) states a
duplicate is admissible to the same extent as an original unless (1) a genuine q
uestion is raised as to the authenticity of the original or (2) in the circumsta
nces it would be unfair to admit the duplicate in lieu of the original. This mea
ns a photograph can be stored digitally in a computer, that a digital photograph
stored in a computer is considered an original, and any exact copy of the digit
al photograph is admissible as evidence.
The State Level
Check your state's rules of evidence for specifics on the admissibility of digit
al photographs. Most states have laws that apply to digital evidence.
As an example, California Evidence Code Section 1500.6(a) (Admissibility of Prin
ted Representation of Images Stored on Video or Digital Media to Prove Existence
and Content of Image) states a printed representation of an image stored on vid
eo or digital media shall be admissible to prove the existence and content of th
e image stored on the video or digital media. Images stored on video or digital
media, or copies of images stored on video or digital media, shall not be render
ed inadmissible by the best evidence rule. Printed representation of images stor
ed on video or digital media shall be presumed to be accurate representations of
the images that they purport to represent.
Photographs as Evidence
The principal requirements to admit a photograph (digital or film-based) into ev
idence are relevance and authentication. Unless the photograph is admitted by th
e stipulation of both parties, the party attempting to admit the photograph into
evidence must be prepared to offer testimony that the photograph is an accurate
representation of the scene. This usually means someone must testify that the p
hotograph accurately portrays the scene as viewed by that witness.
Guidelines for Ensuring Your Digital Photographs Are Admissible
Develop a Standard Operating Procedure (SOP), Department Policy, or General Orde
r on the use of digital imaging. The SOP should include when digital imaging is
used, chain of custody, image security, image enhancement, and release and avail
ability of digital images. The SOP should not apply just to digital, but should
also include film-based and video applications as well.
Most importantly, preserve the original digital image. This can be done a variet
y of ways including saving the image file to a hard drive or recording the image
file to a CD. Some agencies elect to use image security software.
Digital images should be preserved in their original file formats. The saving of
a file in some file formats subject the image to lossy compression. If lossy co
mpression is used critical image information may be lost and artifacts introduce
d as a result of the compression process.
If images are stored on a computer workstation or server, and several individual
s would have access to the image files, make the files read-only for all but you
r evidence or photo lab staff. As an example, detectives could view any image fi
les but they would not have rights to delete or overwrite those files.
If an image is to be analyzed or enhanced the new image files created should be
saved as new file names. The original file must not be replaced (overwritten) wi
th a new file.
Check with Your Legal Advisor
When beginning a new procedure for collecting evidence or recording a crime scen
e, it is always prudent to check with your legal advisor. Consider the Federal R
ules of Evidence, your state's rules of evidence, and other court decisions. Two
court decisions regarding digital images include:
State of Washington vs. Eric Hayden, 1995: A homicide case was taken through a K
elly-Frye hearing in which the defense specifically objected on the grounds that
the digital images were manipulated. The court authorized the use of digital im
aging and the defendant was found guilty. In 1998 the Appellate Court upheld the
case on appeal.
State of California vs. Phillip Lee Jackson, 1995: The San Diego (CA) Police Dep
artment used digital image processing on a fingerprint in a double homicide case
. The defense asked for a Kelly-Frye hearing, but the court ruled this unnecessa
ry on the argument that digital processing is a readily accepted practice in for
ensics and that new information was not added to the image.
About the Author
Steven Staggs is a forensic photography Instructor and has instructed over 3,000
crime scene technicians and detectives in crime scene and evidence photography
over the past 24 years. He is a retired police manager having been in law enforc
ement for 32 years and is the author of the book, Crime Scene and Evidence Photo
grapher's Guide.
Admissibility of digital images as evidence
August 2005, CCTV, Surveillance
A common misconception in the surveillance and monitoring industry is that digit
al images are inadmissible as evidence in court. This white paper examines the c
urrent law in the UK on the admissibility of digital images and concludes that s
uch images are indeed admissible. In so doing, this white paper also briefly exa
mines the techniques used in video compression and highlights current 'best prac
tice' in digital image authentication. This white paper is particularly timely i
n that the Police Scientific Development Branch has just published up-to-date gu
idance on the use of digital images as evidence1.
Introduction
Digital technology is ubiquitous. For years we have listened to digital recordin
gs of music on CDs and generally accepted that the quality is superior to that c
ontained on a vinyl record or tape cassette. More recently we have begun to list
en to digital radio and have benefited from improved sound quality and increased
content. Many of us now watch digital television and take our holiday snaps wit
h a digital camera. The trend is clear; namely, that as computer processing powe
r increases and becomes cheaper, old analog technology is being replaced by new
improved digital systems.
This trend is also now evident in the surveillance and monitoring industry. Incr
easing numbers of CCTV equipment manufacturers are producing digital products (s
uch as digital video recorders (DVRs) and networked video cameras)2. These CCTV
equipment manufacturers recognise the significant advantages that digital techno
logy has over analog technology3.
However, some in the surveillance and monitoring industry still regard digital C
CTV equipment with suspicion. This is perfectly understandable in that the surve
illance and monitoring industry has only recently started to migrate across to d
igital technology. As with all new technologies there is a certain 'fear of the
unknown'. One such perceived 'unknown' is whether or not CCTV images captured di
gitally will be admissible as evidence in court. This white paper summarises the
law in the UK on this particular point4.
What is a digital image?
Before we examine the law on the admissibility of digital images as evidence in
UK courts, we should clarify what is meant in this white paper by the phrase 'di
gital image'5. For present purposes a 'digital image' is either a digital still
photograph or a sequence of digital video which can be viewed on a suitable disp
lay medium (including, for example, a computer monitor, a television, a CCTV mon
itor or a print out). A photograph or sequence of video is digital if it is stor
ed in binary format in memory (including, for example, DVDs, CD-Roms, diskettes,
RAM, hard drives, RAID Arrays and tape streamer systems). Digital images can be
contrasted with more familiar analog images which are typically captured as a v
ariation in some physical property of the recording medium (including, for examp
le, VHS cassettes or photographic film).
The binary nature of digital images gives them certain advantages over analog im
ages. For example, digital images are generally easier to copy perfectly than an
alog images. Moreover, because a digital image is simply recorded as a set of da
ta in computer memory, it is generally easier to process a digital image than an
analog one.
Such processing could, for example, include image enhancement or image analysis.
It could also include image manipulation. In addition, before a digital image i
s even stored in memory it is likely to have undergone some processing. Often su
ch processing will involve the compression of the digital image in order to save
memory or to make the digital image easier to transmit across a network onto a
storage/viewing device. A common example of this compression processing is the D
VD. This shows that digital image compression does not necessarily impair image
quality.
It is, however, exactly this ease with which digital images can be processed, or
are processed before being recorded in memory, which has caused some in the sur
veillance and monitoring industry to doubt whether or not digital images are adm
issible as evidence in court. There are two principal concerns here. The first i
s that often a digital image will have undergone some compression processing bef
ore being stored in memory. The second is that digital images are relatively eas
y to manipulate.
What is compression processing?
In order to assess how real this first concern is, a high-level understanding of
video compression processing is required.
The twin objectives of video compression processing are to reduce the amount of
data required to store/send a digital image whilst also maintaining the quality
of that digital image. International standards bodies such as the ITU-T and the
ISO6 formulate and publish various compression processing standards. There are t
wo main ISO subdivisions that define and regulate image compression standards: n
amely, JPEG and MPEG. JPEG - Joint Photographic Experts Group - is a group of sc
ientists and industrialists who collectively define and regulate standards for t
he compression of still images (including, for example, photographs and individu
al frames captured from video footage). MPEG - Motion Picture Experts Group - un
dertakes the same activities in respect of motion video (that is, two or more st
ill images that together form a piece of video footage). These two bodies have s
et standards such as JPEG (commonly used in digital cameras) and MPEG-2 (the cur
rent DVD standard). ITU-T (a telecoms focused standards body) has also published
standards, such as H.261 and H.263, which define how to compress motion video.
However, whereas MPEG-2 is targeted at the digital broadcast market, H.26x (a co
llective term for the two ITU-T standards) is intended for the compression of vi
deo down to a level that can be transmitted across telephone lines and computer
data networks.
To compress an image using the JPEG standard, the following steps are required:
* Image capture.
* Image transformation.
* Quantisation.
* Entropy encoding.
To compress a video sequence in accordance with one of the H.26x standards or on
e of the MPEG standards (commonly known as motion video compression) requires th
e addition of one more step - namely, motion compensation - resulting in the fol
lowing process:
* Image capture.
* Image transform.
* Motion compensation.
* Quantisation.
* Entropy encoding.
Image capture
At this stage in compression processing, the video compression device (typically
referred to as an encoder7) receives the image from the lens or network, conver
ts it from an analog signal into digital format and stores it into some kind of
memory that the later compression processing stages can then use. This 'incoming
' image may be down-sampled to a lower resolution to reduce the work required by
the later compression stages, thus drastically reducing the amount of data like
ly to be produced at the end of the compression process. This down-sampling is o
ne of the most important stages in video compression, and is known as a 'lossy'
stage (meaning that the data that is discarded in the down-scaling process canno
t be reproduced perfectly).
Image transformation
The next stage in the compression process is for the digital image to be convert
ed into a form that is easily understood by the encoder and by the mathematics t
hat perform the compression. Accordingly, this stage simply involves a translati
on of the image data into a different format. This stage is lossless (meaning th
at data can be accurately transformed between the two formats without loss of in
formation).
Quantisation
This is the stage at which a great deal of compression occurs. The data produced
by the image transform stage is analysed and certain mathematics are performed
on it in order to 'shrink' it. Generally this 'shrinking' process is lossless. T
here are, however, some parts of this 'shrinking' process which are lossy. Both
JPEG and MPEG have performed much research into the sensitivity of the human vis
ual system and have identified that human eyes and brains are more sensitive to
low frequency light than to that at higher frequencies. That being so, quantisat
ion actually reduces the signal strength of the higher frequencies much more tha
n it does the lower frequency signals.
In so doing, some of the higher frequency light signals are lost. In practical t
erms this means that those parts of an image which humans cannot or struggle to
see are discarded during quantisation. This part of the quantisation process is
lossy.
Entropy encoding
Entropy encoding is basically a method of compression that has been utilised in
many information storage and transmission systems including computerised data va
ults and even Morse code. To perform true entropy encoding, the message that is
to be transmitted or stored is first analysed to detect how often each piece of
information (a datum) actually occurs in a message. Each datum is then given a c
ode, the length of which depends on how frequently that datum occurs in the mess
age. For example, if this paper was to be entropy encoded, the letter 'E' would
be given the shortest code, as it occurs the most frequently. If the letter 'Z'
occurs at all (it does now) then it would be given the longest code. Accordingly
, this entire article (and in fact any English language text) could be compresse
d down to a fraction of its original size. A similar process takes place in the
entropy encoding of a digital image. The key point to note is that the entropy e
ncoding of a digital image is lossless.
Motion compensation
This stage is included only in motion video compression standards (such as H.261
and MPEG). Motion compensation is also known as 'motion estimation' or 'motion
prediction' - although these terms are perhaps a little misleading in that no es
timation or prediction actually takes place. The objective of this step is to ex
amine the image to be compressed (the encoder image) and compare it to the previ
ously compressed image (the reference image). Motion compensators look for: (i)
new elements in the encoder image; (ii) moved elements in the encoder image; (ii
i) changed elements in the encoder image; and (iv) unchanged elements between th
e encoder and reference images. Once the encoder has analysed the two images, lo
oking for these specific elements, the rest of the encoding process is defined:
* New elements are quantised, then entropy encoded.
* The motion of moved elements is encoded.
* The changes in elements are either encoded, or if the change is too great, the
element is re-quantised and then entropy encoded.
* Unchanged elements are refreshed at regular intervals, to prevent 'drift' betw
een the compressed and uncompressed images.
This is a high-level summary of the operation of motion compensators which gloss
es over many of their complexities. It does however demonstrate that the princip
al function of motion compensation is one of information management. The degree
of lossiness of the motion compensation stage of an encoder will depend on how a
ccurately that encoder is set to 'manage' the digital image information; the les
s accurate the 'management' of the digital image information, the more lossy the
motion compensation stage of encoding will be. Typically, encoders enable the u
ser to set the level of accuracy of information management and accordingly the l
evel of lossiness of the motion compensation stage.
The additional step of motion compensation included in standards such as H.263 a
nd MPEG allows motion video sequences to be compressed far more effectively than
by compressing each individual frame in a video sequence into a JPEG image - th
is latter option is commonly known as MJPEG. The difference in compression ratio
s between MJPEG and H.263 encoded sequences can be greater than 600%. This is a
huge efficiency gain and results in video recordings that maintain the fidelity
of motion but reduce the bandwidth and storage requirements of the system by up
to six times.
Another way to look at the benefits of motion video compression is to do a like
for like comparison, as shown in the following table.
The foregoing summary of image compression processing highlights that image info
rmation is only discarded at three stages in the process; namely at image captur
e, at quantisation and, possibly also at motion compensation. Moreover, the amou
nt of information that is discarded may often be controlled by the user of the e
ncoder. In other words, most encoders enable the user to set the quality of the
digital image to be recorded; generally, the better the quality of the digital i
mage, the less information is discarded during compression.
It should now be apparent that the compression of digital images is actually qui
te a procedurally straightforward (if mathematically complex and intensive) exer
cise. This was certainly the view reached by the House of Lords Science and Tech
nology Select Committee in its report entitled 'Digital images as evidence'8. To
date this remains the most authoritative report on the admissibility of digital
images as evidence in the UK courts. The Committee examined the various rules9
on the admissibility of images as evidence and came to the clear conclusion that
:
"Digital images, which we initially thought might create difficulties for the co
urts, do not. But many people think that there will be difficulty in obtaining l
egal acceptance of digital images."10
In other words, the Committee came to the general conclusion that digital images
are admissible as evidence in both civil and criminal cases before the UK court
s, provided however that, like other types of evidence, those digital images are
appropriately authenticated.