This article was downloaded by: [UQ Library]

On: 05 November 2014, At: 20:35

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Australian Journal of Forensic Sciences

Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/tajf20

CCTV images as evidence

Glenn Porter
Published online: 10 Jun 2009.

To cite this article: Glenn Porter (2009) CCTV images as evidence, Australian Journal of Forensic
Sciences, 41:1, 11-25, DOI: 10.1080/00450610802537960

To link to this article: http://dx.doi.org/10.1080/00450610802537960

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
 Australian Journal of Forensic Sciences
Vol. 41, No. 1, June 2009, 11–25

PLENARY
CCTV images as evidence
Glenn Porter*

Introduction
The introduction of closed-circuit television (CCTV) into the public landscape has
provided a new level of surveillance within contemporary society. Never before
have vast numbers of individuals been under this level of surveillance and scrutiny.
The density of surveillance cameras has increased exponentially in recent years,
Downloaded by [UQ Library] at 20:35 05 November 2014

and other technologies such as cameras in mobile phones provide an itinerant

source of surveillance. This new social phenomenon has raised ethical issues
regarding privacy and several social debates have arisen. Some issues include the
effectiveness of the technology to increase security over privacy, whether the
security results in better or real protection of citizens, and whether the technology
can be used as an effective crime deterrent tool.
The number of CCTV cameras situated in communities is, in most cases, an
unknown quantity. Some sources have suggested there are over 4 million CCTV
cameras installed in the UK1. A report for the Australian Research Council
(ARC) by Helene Wells, Troy Allard and Paul Wilson from Bond University
examined crime and CCTV in Australia2. The study was more specifically focused
on the Gold Coast (Queensland) and the Queensland Rail (QR) Cityrail network
and was not Australia-wide. The report suggests that the original objective for
installing CCTV was to reduce crime by addressing alcohol-related violence and
anti-social behaviour in an attempt to improve public safety2. It also reported on
the density of CCTV cameras installed by Queensland Rail at railway stations and
car parks.
The Wells, Allard and Wilson report states there were a total of 283 CCTV cameras
installed by QR in 1995 and by 2003, that number had grown to 3369 cameras. This is
an increase of over 1000% in 8 years. Although the data is a little out of date, they
certainly illustrate the trend during that time period. Wells et al. also noted that by 2006
there were 3398 cameras installed on Queensland Rail stations and car parks, which is
only a slight increase from 2003. However, the report further suggests that the inclusion
of cameras installed in train carriages has brought the total number of CCTV cameras
in late 2006 to approximately 5500 (see Figure 1).
There is little doubt that the application of CCTV has grown enormously over
the last decade. Today we see CCTV installed in streets, ATMs, public transport
(trains, buses and taxis), shopping centres, hotels, universities, public buildings and
even private homes. The amount of times we are now recorded by CCTV during our
daily travels is too staggering to consider.

*Email: g.porter@uws.edu.au

ISSN 0045-0618 print/ISSN 1834-562X online

Ó 2009 Australian Academy of Forensic Sciences
DOI: 10.1080/00450610802537960
http://www.informaworld.com
 12 G. Porter
Downloaded by [UQ Library] at 20:35 05 November 2014

Figure 1. Number of CCTV cameras installed by QR Citynail network [Data sourced from
Wells et al. 2006].

It has been suggested that the intrusion into our privacy is off-set by the increased
security CCTV provides to citizens while transiting through public spaces. This
discussion will not examine the civil liberty debates associated with privacy issues of
law abiding people; it does accept, however, that CCTV captures criminal activity in
action. It is the reliability of this imagery to provide forensic evidence that is the
focus of this discussion.

The effectiveness of CCTV

Recent media reports have suggested that the effectiveness of CCTV to reduce crime
in the UK is under serious doubt. Detective Inspector Mike Neville, who leads
Scotland Yard’s ‘Visual Images, Identifications and Detections Office’ is reported to
have said at a recent conference that only 3% of London’s robberies had been solved
using CCTV images. He further suggests that no thought has gone into how police
should use CCTV images and that it is an ‘utter fiasco’3–5. The Wells, Allard and
Wilson report also concluded that ‘the effectiveness of CCTV as a crime prevention
tool is questionable’; however, it appeared effective at detecting violent crimes2. I
do not find these results or attitudes surprising. There has been little research into
how CCTV images may be used effectively as forensic evidence or aspects of
reliability.

CCTV as evidence
Most Western adversarial legal systems have established rules of evidence. General
principles regarding the admissibility of photographic evidence may be summarised
into three general standards and apply to most legal jurisdictions. Photographic
evidence may be defined as: photographs, digital images, video footage, composite
digital images or computer drawings (identikit images), CCTV images and images
taken by mobile phones. May (1987) suggests the photographs or images become a
 Australian Journal of Forensic Sciences 13

tangible form of evidence6. The three general standards that apply to photographic
evidence include:

(1) the photographic evidence must be ‘relevant’;

(2) the photographic evidence must be ‘authenticated’; and
(3) the photographic evidence must not be ‘unfairly prejudicial’.

These standards certainly apply to CCTV images. The application of CCTV images
during the investigation of a crime is certainly multifaceted and produces a passive
form of witnessing. CCTV images can:

(1) provide a visual narrative of the event for interpretation;

(2) provide visual intelligence of physical aspects depicted in the images;
(3) provide identification;
(4) provide exhibits for court; and
Downloaded by [UQ Library] at 20:35 05 November 2014

(5) provide public safety by the detection of criminal activity (if monitored).

Forensic examination activities associated with CCTV images as evidence include:

(1) interpretation;
(2) analysis of image data and reconstruction;
(3) image enhancement;
(4) image authentication; and
(5) facial identification.

I will attempt to look at these five activities in more detail.

Interpretation
The interpretation of visual narratives from CCTV images may be in several cases a
simplistic representation of an event (for example, someone vandalising a bus shelter
by smashing the glass). However, other events present highly complex situations that
are more difficult to interpret. The danger in interrupting visual evidence is that the
complexity is often concealed in the simplistic representation of the images. A good
example to reflect on is the ‘Rodney King Video’, where both sides of the legal
representation presented completely different interpretations of the same video
footage7,8. The defence in the case against the police officers also suggested that the
video images, when viewed from a simple representational perspective, actually
distorted the reality7,8.
Janina Struk makes an observation regarding interpreting the vast collection of
photographic evidence from the Holocaust during the Second World War. In her
book entitled Photographing the Holocaust: Interpretations of the Evidence; Struk
suggests:
The photographic representation of the Holocaust does not give a comprehensive
account of the historical events which photographic narratives generally lead us to
believe; that is not possible. Photographs are fragments. They illustrate stories, they do
not tell them. It has been left to the curators, film-makers, historians and propagandists
to determine how they are interpreted.9
 14 G. Porter

How we interpret CCTV images as evidence is also more complex than what it may
appear. Establishing an understanding of what is ‘real or distorted’, ‘true or untrue’
requires an understanding of visual culture aspects that relate directly to forensic
science. Concepts regarding ‘photographic truth’ are currently being discussed
quite intensely among visual culture academics. Its relevance to forensic science
and visual evidence needs to be further explored. The danger when inappropriate
and inaccurate interpretations are made before a lay jury is the weighting photo-
graphic evidence may have against the accused. The highly believable nature of
photographic evidence may present an unfair prejudice that could be most difficult to
defend against.
Western societies are considered to have a high level of visual literacy. Our lives
are saturated with the application of images and we all use skills to interpret the
messages associated with images in the media, advertising, signs and other aspects
of our lives. How skilled are we however, when required to interpret images as
evidence?
Downloaded by [UQ Library] at 20:35 05 November 2014

Well-known Dutch artist Mauritis Cornelis (M.C.) Escher’s (1898–1972)

woodcut lithographs are famous for not only for the artistic body of work10,11,
but for their illusionary aspects which have been commercialised numerous times
over. Escher’s work produces buildings that are constructed with impossible
perspectives that illustrate a stairwell that continually travels up or continually
down, depending on the direction of the gaze (‘Ascending and Descending’ 1960).
Three of his works that illustrate this impossible perspective include: ‘Belvedere’
(1958), ‘Ascending and Descending’ (1960) and ‘Waterfall’ (1961)11.
Escher’s constructions form the basis of an ‘impossible triangle’ (see Figure 2)
which was consider to be first illustrated by a Swedish artist Oscar Reutersvard
(1915–2002) in 1934. Reutersvard’s work represented impossible figures and his
impossible triangle was actually a series of cubes to form a triangular shape. The
geometric structure of the impossible triangle was first articulated and popularised
by Roger and Lionel Penrose in 1958 in an article in the British Journal of
Psychology and is also known as the ‘Penrose triangle’11–13.
The drawing in Figure 2 depicts the structure of an impossible triangle as
described by Penrose and Penrose in 1958. It is called an impossible triangle because
the perspective cannot exist in real life. If the item could not exist in real life, then it
would be fair to assume obtaining a straight photograph (without digital retouching)

Figure 2. An impossible Triangle aka the Penrose Triangle.

 Australian Journal of Forensic Sciences 15

would not be possible. This thinking suggests that an item must exist in reality before
it may be photographed.
Figure 3 is, however, a photograph of an impossible triangle, and it has been
taken with a camera with no digital manipulation whatsoever. It is a straight
photograph of the timber structure. If so, then how could this structure exist in real
life: did it or didn’t it?
Downloaded by [UQ Library] at 20:35 05 November 2014

Figure 3. A photograph depicting an impossible triangle. Photograph by Glenn Porter.

We certainly do interpret this object as an impossible triangle, but it is not what it

seems. You cannot photograph objects that do not exist; however, the interpreta-
tions may be quite inaccurate based on the reality. The solution to this photographic
puzzle is revealed in Figure 4. The misinterpretation is actually due to some strong
Gestalt psychology principles including continuity, similarity and proximity13.
Another question regarding the interpretation of CCTV images and other visual
evidence is: who should make these interpretations? Should the images be interpreted
by an expert with training, knowledge and experience in visual culture, or can any

Figure 4. Photographs depicting an impossible triangle. Photographs by Glenn Porter.

 16 G. Porter

person make an interpretation? Should the courts provide a caution to the jury when
interpretation of visual evidence is presented? Should investigators be provided with
training to provide informed interpretations? In many instances, understanding what
you cannot say is usually more important than what you say. These are certainly
perplexing questions on a form of evidence that can be dangerously taken for
granted, particularly CCTV images.

Analysis and reconstruction

On a more positive note, CCTV images can provide valuable intelligence by
examining the physical attributes presented as visual evidence. Physical attributes
may include the colour and type of clothing, the height of the person of interest, the
time of the event, the chronology of the event, the number of offenders and several
others. These attributes may be counted, timed, measured and can be very useful
when reconstructing aspects of the visual evidence.
Downloaded by [UQ Library] at 20:35 05 November 2014

CCTV systems also provide metadata to be embedded into the file that record
time. Fixed landmarks depicted in the scene also allow distance and scale to be
measured. In one case, we were able to estimate the average speed of a truck using
quite visible landmarks on the road. The time presented in the metadata found in the
digital imaging files and the measured distance between the landmarks at the scene
were used to calculate the average speed7. The reconstruction of the scene using
physical attributes is certainly an aspect worth exploring with CCTV images. As a
cautionary note, however, these attributes require careful consideration of
photographic concepts like perspective distortion. I will elaborate in more detail
on this aspect during the discussion on identification.

Image enhancement and image quality

Poorly resolved images are more common than not with current CCTV technology. The
Home Office Scientific Development Branch in the UK published an operational
manual for CCTV in 200714. There are significant gaps between the operational aspects
of CCTV technology and the requirements needed for forensic investigation.
The authors of the Home Office publication identify four image quality
parameters14:

(1) clarity – overall sharpness;

(2) detail – the ability of the image to resolve fine detail;
(3) colour – whether the colour is natural due to the light source and calibration
of the camera; and
(4) artefacts – image artefacts include excessive noise, lens flare, lens distortion
and other faults.

Image enhancement is a process that provides improvement to the image quality.

There are several software applications that can perform image enhancement. The
latest version of Adobe PhotoshopTM (CS3) has included some new tools for
enhancement of video images including a de-interlacing filter that separates over-
lapping video images15. Other highly useful applications include: noise reduction,
sharpening and lens distortion correction. Naturally, a quality image in the first
instance is always preferable to a poor image that has been enhanced.
 Australian Journal of Forensic Sciences 17

Poor image quality is a significant problem with CCTV images and particularly
digital CCTV. Factors influencing image quality include lens optics, lighting
conditions, exposure, sensor size and the digital file storage application. A significant
problem with digital systems is how to store the vast amounts of memory needed for
each CCTV camera. The memory requirements notably increase with higher
resolution and higher frame rates. To reduce the amount of memory needed for the
storage of digital CCTV images, two strategies are employed:

(1) smaller frame-rates are used. For example, 24–30 frames per second are
required for natural motion. CCTV cameras can capture images as low as
2–4 frames per second, which results in less images, therefore less storage;
(2) codec algorithms are also employed. These are digital compression and
decompression processes that allow the stored image to be reduced in sized
and decompressed when played using a player that can read the codec14. MP3
format is an example of a codec.
Downloaded by [UQ Library] at 20:35 05 November 2014

Codecs are highly problematic for CCTV images used as evidence. Quality is
significantly reduced during the compression and data is lost. Another major
problem with this system is that codec processes are not generic and are specific to
the system used to record the images. Digital photography also had this problem
with RAW files; however, industry standards have now included standard RAW
processing applications including Adobe RawTM and generic DNGTM (Digital
Negative) file applications.
Codec algorithms also greatly reduce the ability to download highly resolved still
images for forensic analysis. Instead, the forensic practitioner is forced to use the
propriety software player that supports the file codec and CCTV system. This is a
significant problem with CCTV images as evidence, because the general player’s
ability to capture still frames is of a low standard.

Image authentication
The provenance of CCTV images, and any other images, is an important
consideration that is often overlooked. While I do not personally know of any
incidents of falsified CCTV images, there is a long history of faked photography used
mostly for propaganda16,17.
With the increasing application of image-editing software, the likelihood of falsified
images being presented to police is ever-increasing. Despite a common myth, altered
images are quite readily detectable by forensic photography experts. Inconsistencies in
image noise, lighting direction, colour, vanishing points, size, image perspective are just
a few aspects that may lead to the detection of faked images18.

Identification
Identification of individuals depicted in CCTV images offers a lot of promise for
forensic intelligence and evidence. There is an important distinction between
‘recognition’ and ‘identification’. Recognition provides some visual similarities
between the facial features of the subject in the image and the facial features of a
known subject. Recognition may be achieved by someone who knows a person
that presents some similarities in the images.
 18 G. Porter

Identification requires a comparative examination between the questioned image

(CCTV source) and an exemplar (known) image. The examination is based on a
concept of individuality or uniqueness of facial features. Understanding what
constitutes individual qualities is an area of some contention. Recent research
findings from Canada have suggested that there is little variation between individuals
based on facial morphology. The Will and William West incident in the United
States in 1903 where two prisoners of the Leavenworth Penitentiary had almost
identical Bertillon anthropometric results proved highly problematic for the
Bertillon identification method19.
There is some common agreement that unique identifiers may include facial
imperfections such as scars, moles and freckles, etc. It is suggested these markings
and their orientation are random enough to be considered as unique. The
morphology of the ear is also considered as a unique feature between individuals,
like fingerprints.
A method of facial identification was published in 2000 by Porter and Doran that
Downloaded by [UQ Library] at 20:35 05 November 2014

suggested a holistic approach that combined unique identifying features in

corroboration with other facial aspects. Porter and Doran suggested a holistic
method should consider all of the following aspects during the examination20:

(1) individual facial characteristics (moles, freckles, scars, etc.);

(2) form, size and shape of facial features (facial morphology);
(3) facial symmetry (or degree of asymmetry); and
(4) anthropometric analysis.

The proposed holistic facial identification model suggests several approaches be used
to corroborate the results. However, it further supports that without any evidence of
unique identifying features, a positive identification would be most inappropriate.
The difficulties currently seen with most CCTV images is their inability to resolve
fine detail that could provide the visualisation of any facial features that may be
considered as unique identifiers. Poorly resolved CCTV images prohibit effective
examinations.
Some recent Australian cases have produced positive identifications based on
facial morphological aspects and without any evidence of unique characteristics.
These cases used CCTV images of poor image quality which would not be able to
resolve any unique features if they were present. Positive identifications should be
also backed-up with a sound method of articulating individuality.
Establishing and testing concepts involving individualisation is an area of
research that needs considerable attention for CCTV evidence. Further research may
provide better tools for this type of evidence. Whether definitive positive
identifications are possible with CCTV images will largely depend on the resolving
power of the imaging system and the features presented by the subject. At this stage,
definitive positive identifications are most unlikely and, in most cases, dangerous.

The influence of image perspective on facial morphology

A question arose from some casework involving facial identification from CCTV
images that made a positive identification based on facial morphology without any
unique features. The identification method involved a comparison of numerous
morphological aspects including mostly size of facial features. The examination
 Australian Journal of Forensic Sciences 19

involved the comparison of images taken from a CCTV camera situated in an ATM
and exemplar material was taken of the accused who was in custody.
My concerns were regarding the consistency of the size of features (the morpho-
logy) between images displaying different image perspective. Image perspective
changes with ‘u’ distances (distance between the subject and lens). I was able to
establish through the police photographer that the exemplar photographs were taken
at a distance between 3 and 4 m. Distance ranges from the CCTV images were
measured by an examination of the scene and using several landmarks to establish
the points of interest.
The ‘u’ distances of the CCTV image was estimated between 0.4 and 1 m, while
the exemplar images were taken at a distance of 3–4 m. From a photographic aspect,
the different ‘u’ distances may produce significant variation in the representation of
facial morphology between the questioned and exemplar photographs due to the
different inherent image perspectives. Image perspective is a complex visual
phenomenon that is not well understood outside the photography discipline,
Downloaded by [UQ Library] at 20:35 05 November 2014

including forensic practitioners. It results from the transitional process of three-

dimensional objects when made into two dimensions (photographs). Another
exemplar image was taken during the arrest of the accused and this ‘u’ distance was
not established. The question raised in this case was if facial morphology was the
main form of the identification evidence, the photographs used in the comparative
examination should really be of similar image perspective. Otherwise, the
comparison between size of facial features could be erroneous.
An experiment was designed to test and measure the variations in facial
morphology that may result with images displaying dissimilar image perspective.
Image perspective is controlled by the ‘u’ distance as an independent variable. The
experiment attempted to answer two questions.

. Does significant variance occur in facial morphology when using different

image perspective?
. Is there an equilibrium point where changes in image perspective no longer
alter facial morphology?

The experiment involved a model photographed with different sized facial features
at different distances. A visual analysis was first conducted followed by photo-
anthropometric analysis of anatomical landmarks. The results indicated significant
variance in facial morphology indices at distances 51.5 m, moderate variation at
distances between 1.5 and 3 m, and insignificant variation at distances 43 m.
This experiment indicated that there was a critical need to examine questioned
and exemplar images that were of similar image perspective. This means that when
photographs are taken of Persons of Interest (POI) for comparison, the ‘u’ distance
of the questioned CCTV images must first be determined. This involves examining
the scene and using various photographic or photogrammetry techniques to
establish these distances. Once the questioned images ‘u’ distances are determined,
the exemplar photographs should then be taken at the same distance to produce an
image of similar perspective. Naturally, the camera viewpoint in relation to the
CCTV camera and the POI should also be replicated. Camera viewpoint and image
perspective are two very different photographic conditions and should not be
confused. It makes sense to approach these comparative analysis methods for facial
identification with like-to-like specimens.
 20 G. Porter

Figure 5 provides a visual analysis of the variances in facial morphology due to

different ‘u’ distances. Note the relationship of size between the ear and nose in each
image. Figure 6 illustrates the photoanthropometric indices differences between the
known values measured from the left and right alare – left and right lateral canthus
and the indices values measured from the photographs taken at different ‘u’
distances. Significant variation is noted at the shorter ‘u’ distances.
Downloaded by [UQ Library] at 20:35 05 November 2014

Figure 5. Visual analysis of variation of facial morphology caused by different camera ‘u’
distances. Photographs by Glenn Porter.

Figure 6. Variation of photoanthropometric indices of known values and indices values

measured from photographs.
 Australian Journal of Forensic Sciences 21

Estimation of POI height

Another research area we are trying to gain some understanding of at UWS is
calculating the error rate regarding determining the height of POIs captured on
CCTV against their known height. The method we are applying is a simplistic one
that involves overlaying images taken with the CCTV camera with references used
to:

(1) determining the location of the POI within the image space using a technique
called ‘perspective grid photogrammetry’21–23; and
(2) apply a surveyor’s scale positioned on the location of the POI.

The technique uses Adobe Photoshop CS3TM to overlay and register the images. The
critical component to this technique is the application of the same camera that captured
the event or POI being used also for the analysis. The CCTV must be a static camera
Downloaded by [UQ Library] at 20:35 05 November 2014

and not a camera that can pan, tilt and zoom. The advantage of this technique is
its simplicity and the fact that the optical system remains constant, meaning the same
lens focal length, focus and other optical conditions are always used.
The question this work is attempting to answer is whether the error rate for
estimating height is smaller than the height range from a normal population. If
the error rate is the same or larger than the range calculated from a sample
population, then the application of height estimation becomes meaningless. If
found suitable, however, which is highly questionable, the estimation of height is
an application that would not be suitable for identification purposes in any event.
However, it may have some application for the elimination of POIs, and this
would certainly be a positive contribution.
Figure 7 is an experimental image that combines the CCTV image of the POI
with an image taken by the same camera with a surveyors scale placed at the location
of the POI. The opacity of the image with the scale was reduced to allow the POI
image to be seen. Note the degree of registration of most objects in the scene (pavers,
bricks, notice board) with the exception of some notices on the board due to
movement. This image also shows the difficulties with this type of analysis. The

Figure 7. Experimental images combining the CCTV image of a volunteer POI with an
image with a surveyor’s scale to measure the POI height.
 22 G. Porter

posture of the POI would provide a significant degree of error. The selection of
images is an important consideration with this technique.
Determining the placement of the surveyor’s scale requires calculation using
perspective grid photogrammetry. Perspective grid photogrammetry is a single image
photogrammetry technique (as opposed to stereo images) and requires a further
photograph taken with the same CCTV camera and with the inclusion of two (or
four) square grid references21–23. These references provide the following visual
information needed for the analysis:

(1) known size or scale;

(2) known location within the scene/image; and
(3) indicate the degree of perspective distortion and determines the required
vanishing points.

The grid references are placed within the scene and the location noted using
Downloaded by [UQ Library] at 20:35 05 November 2014

standard crime scene triangulation methods. This location is needed as a standard

to measure from once the location of the POI has been determined by the
perspective grid photogrammetry. The image taken with the grid references are
combined with the POI CCTV image and the photogrammetry applications
applied. Any inaccuracies are quickly discovered by the image registration on the
combined (overlaid) image. Figure 8 demonstrates this application made directly
on the combined image. Reverse projection photogrammetry is another method
that may be used to determine the location of the POI23.

Figure 8. Location of the POI determined using perspective grid photogrammetry.

 Australian Journal of Forensic Sciences 23

Establishing minimum resolution standards for unique identifiers

Another research project we are working on at UWS is determining the required
minimum resolution value for CCTV cameras to resolve the morphology of an ear.
The experimental design uses the ear of subjects as a reference because it is
considered the largest unique identifier source position on the head. This work is
currently being undertaken by Honours student Lisa Cotterill.
Experimental images were captured at various ranges and a test target is also
recorded using the same imaging system and optical parameters. These images were
then treated to reduce the level of resolution using digital editing software called
Focus MagicTM which provides alterations to the image focus by a controlled
‘defocusing’. A series of images at differing degrees of resolution were produced from
each range. The range of test images of various POI volunteers were taken at
established distance ranges that depict certain percentages of the camera frame.
Figure 9 is a sample of one sequence (not complete) that shows the test target
embedded into the image and the image resolution reduced using Focus MagicTM
Downloaded by [UQ Library] at 20:35 05 November 2014

software. Thresholds regarding the visibility of the ear morphology are established
under a controlled viewing method. The images that display or maintain a clear
visual rendition of the ear morphology are identified and subjected to resolution
measurements from the test target using Modulated Transfer Functions (MTF).
MTF is a quantitative method of determining resolution across a range of spacial
frequencies unlike resolving power which is highly subjective and the value
represents a single frequency. MTF is traditionally a system of measuring the
performance of an optical system using a sinusoidal test target recorded onto film.
Resolution is determined by measuring the contrast of the recorded test target using
a microdensitometer and the results are plotted in a sinusoidal waveform. As the
contrast reduces, so does the level of resolution or ability to record higher spacial

Figure 9. A series of experimental images taken of a volunteer POI with the test targets
embedded into the image for MTF evaluation.
 24 G. Porter

Figure 10. Sample curves measured from the images and test targets displayed in figure 9.

frequencies. A MTF value of 1 or 100% represents the ideal lens performance. Film-
based optical systems are measured by MTF using wave recordings while digital
systems use edge methods23.
Digital images use a MTF system by measuring a single slant edge component of
the target. An edge profile is first established and the MTF is produced by a Fourier
Downloaded by [UQ Library] at 20:35 05 November 2014

transfer process of the edge profile24. The resulting MTF curve represents the
performance across a range of spacial frequencies. While this research is still in the
developing stages, the MTF curves in Figure 10 are the results from the samples used
in Figure 9. The edge profile is indicated on the top of each box while the MTF
values are displayed in the lower section. These values are only indications at this
stage and the work is ongoing.

Conclusion
As a way of summing up, I’ll make the following recommendations regarding
identification evidence from CCTV images. I suggest we be patient with this form of
evidence and resist making claims about POI identification without the support of
the technology. Improvements in image quality are bound to develop especially if we
can establish some research links between the technology developers and forensic
science practitioners. We need to exercise patience so we do not make expensive
mistakes including possible misidentifications, injustices and produce case law that
may prohibit using this form of evidence in the future. If we can develop research
links with CCTV technology developers, develop proven tools that can assist in the
concepts of individualisation, develop training for practitioners and investigators
who may use the technology as a source of evidence, CCTV images as evidence has a
bright future. It is no surprise that the recent reports from Scotland Yard suggest
CCTV has proven to be ineffective. We simply have not yet developed the concepts,
technology or tools to appropriately support this form of evidence. However, it will
become a tool of immense value once we have.

Acknowledgements
I would like to thank Chester Porter QC for his insightful commentary and my colleague and
friend Charles Crumlish for his assistance with the preparation of the sample images used in
this presentation.

References
1. BBC News. ‘Talking’ CCTV scolds offenders. 2007. 4 April 2007.
2. Wells HA, Allard T, Wilson P. Crime and CCTV in Australia: understanding the
relationship. 2006. Australian Research Council, December 2006.
 Australian Journal of Forensic Sciences 25

3. Bates D. Billions spent on CCTV have failed to cut crime and led to an ‘utter fiasco’, says
Scotland Yard surveillance chief. 2008. Daily Mail, 6 May 2008.
4. Bowcott O. CCTV boom has failed to slash crime, say police. 2008. The Guardian, 6 May
2008.
5. Kerin L. Doubts raised over using CCTV cameras. 2008. ABC News, 7 May 2008.
6. May R. Admissibility of photographic and videotape evidence. Journal of Criminal Law.
1987;51:437–443.
7. Porter G. Visual culture in forensic science. Australian Journal of Forensic Sciences.
2007;39(2):81–92.
8. Sturken M, Cartwright L. Practices of looking: an introduction to visual culture. Oxford:
Oxford University Press; 2003.
9. Struk J. Photographing the Holocaust: interpretations of the evidence. London: I.B.
Tauris; 2005.
10. Escher MC, Locher JL. The infinite world of M.C. Esher. New York: Abradale Press;
1984.
11. Escher MC. The graphic work of M.C. Escher. New York: Gramercy Publishing; 1984.
12. Draper SW. The Penrose triangle and a family of related figures. Perception. 1978;7(3):
283–296.
Downloaded by [UQ Library] at 20:35 05 November 2014

13. Zakia RD. Perception & imaging. 2nd ed. Boston: Focal Press; 2002.
14. Cohen N, Gattuso J, MacLennan-Brown K. CCTV operational requirements manual: is
your CCTV system fit for the purpose? 2007. Version 4, Home Office Scientific
Development Branch, Pub No. 55/06.
15. Reis G. Photoshop CS3 for forensic professionals. Indianapolis: Sybex; 2007.
16. Brugioni DA. Photo fakery: the history and techniques of photographic deception and
manipulation. Virginia: Brassey’s; 1999.
17. Jaubert A. Making people disappear: an amazing chronicle of photographic deception.
Washington: Pergamon-Brassey’s; 1989.
18. Russ JC. Forensic uses of digital imaging. Boca Raton (FL): CRC Press; 2001.
19. Beavan C. Fingerprints: murder and the race to uncover the science of identity. London:
Forth Estate; 2002.
20. Porter G, Doran G. An anatomical and photographic technique for forensic facial
identification. Forensic Science International. 2000;114:97–105.
21. Ray SF. Scientific photography and applied imaging. Melbourne: Focal Press; 1999.
22. Siljander RP. Applied police and fire photography. Springfield (Ill): Charles C Thomas
Publisher; 1976.
23. Robinson EM. Crime scene photography. Amsterdam: Academic Press; 2007.
24. Jacobson RE, Ray SF, Attridge GG, Axford NR. Manual of photography: photographic
and digital imaging. 9th ed. Melbourne: Focal Press; 2005.