For some, debates about the veracity of photographic images rage, others have simply moved on from hackneyed arguments about truth and testimony. However, in police work and forensics, questions about digital photography have not been fully addressed. In the UK, there is no common practice on the capture and storage of digital photographs. The guidelines, such as they are, are open to interpretation by each of the UKʼs police forces, and a key issue centres on whether to shoot raw or JPEG.

It is essential that evidence is permissible in court, so the challenge is to foresee problems with a form of evidence such as digital image data. In time, a form of evidence may be found to be unreliable, and therefore discredited. Science drove the use of DNA profiling and, though science is a broad term, similar rigour has to be applied to the use of digital imaging in police work. There is still the notion that anyone who picks up a DSLR is a photography expert, or that being able to tick a box on a staff development form is a panacea for the complexities of image processing.

The leviathan of the legal system moves at a different pace from the fast-changing landscape of digital imaging; official guidelines can be out of step with operational needs. So the diverse and ambiguous implementation of the ACPO (Association Of Chief Police Officers) digital imaging handling guidelines in 2002 could well lead to convictions being quashed on the technicalities of photographic veracity. Though having some flexibility when interpreting the guidelines can make workflows fit for purpose, each force still has its own Standard Operating Procedures and, as long as those protocols are followed, then all is deemed to be in order. There is a disturbing reliance on JPEG files in several forces, for example, and no matter how secure a system might be, it is extremely difficult to prove that JPEG evidence has not been tampered with. In court, an imaging officer might have to testify that a photograph is a true representation of the captured data and has not been manipulated, yet the court must take their word for it. Immutable evidence is required.

Case for change
As someone involved in advising and training police officers on digital imaging, I advocate the use of raw, because it has obvious advantages over JPEG in forensic work. It is a read-only format, and therefore has integrity in court. JPEG is not wholly wrong, but creating a tamper-proof workflow for JPEG files is impossible, and describing digital workflows to the layperson, such as a juror, would involve esoteric language.

The unprocessed digital data captured by a camera, raw files have a greater tonal latitude than JPEG, which means that processing can draw out shadow detail or pull back highlight information; currently as much as two stops of light. JPEG files are regarded as the print file; the camera has processed the data. As the camera can only ever enhance tone and colour, the JPEG, as written by the camera, has integrity as evidence. But a JPEG file can be processed in Photoshop, and proving where and when enhancements have been made is the fundamental flaw of using a JPEG workflow. There is a potential auditing blind-spot between the camera and the pixel data entering a secure server system.

Enhance and manipulate have similar meanings and often these terms are interchangeable. Without resorting to pedantry, to enhance colour and tone is a benign process, whereas manipulating images can involve the displacement of pixels; pixels can be cloned to remove items from a picture, or place items or people in it. It is this fear of image manipulation that drives the need to prove veracity.

Changes in a digital photograph can be tracked via the metadata embedded with the pixels. So, while fakery has always been part of photography, digital images contain metadata that tracks changes made to photographs. EXIF metadata can record the time, camera model, aperture and shutter speed of every shot taken. It can even record the location of a photograph when GPS modules are used. For many photographers this is common knowledge, but what is less well-known is that the tracking of metadata can extend beyond camera EXIF data, and that editing packages, such as Photoshop, can be enabled, via its built-in History Log, to record when files are opened and what tools were used to enhance or manipulate a photograph.

Although the History Log does not record exactly what has been cloned, it will log that a Healing Brush or Clone Stamp have been used on an image. Combine the time metadata for these edits with the login data for the user at that workstation and you can find out who made changes.

But enough of subterfuge. Best practice requires that an archive image is an exact copy of what the camera recorded on to its original media. So Compactflash cards should be archived to a secure read-only server, making the images stored on the read-only server pixel-perfect duplicates of the captured image. By maintaining an unaltered archive image, a viewer can compare the archived image with the final rendered image. This means that, in court, it can be determined to what extent a photograph has been altered and the degree by which the quality of the photograph has changed.

Raw materials
Any enhancement techniques have to be reproducible. In police work, image-processing techniques need to be repeatable and produce similar results. Here lies a problem for the raw workflow; raw conversion software may not always maintain backwards compatibility with older raw formats. There may come a time that Adobe Camera Raw (ACR) can not open NEFs from a Nikon D1. There may be a solution in Adobeʼs DNG format; however, DNGs are not an exact copy of the original capture data. DNGs log image processing in the History Log just like a native raw file. Such problems can be overcome. Raw Workflows generate the best quality evidence, as well as inherent veracity.

A raw file has a greater bit-depth and tonal latitude than either in-camera JPEGs or TIFFs. Bad exposures can be corrected, tones adjusted and colours balanced to render hues accurately in an almost lossless way, whereas JPEG workflows are destructive, and adjustments made in post-production will delete tones from an image when saving the processed photograph back as JPEG. JPEG files are an end product - the print file. Opening JPEGs to make vital changes to tone or colour in order to draw a courtʼs attention to a significant detail is almost impossible, whereas having raw data provides those working in forensics the best image data possible for their work.

Many images taken in police work are not made by skilled photographers. The person pressing the shutter at a crime scene may not be a specialist photographer, so the chances of getting incorrect exposures are high and, with the police facing cutbacks, this is not likely to be remedied. A raw workflow offers the best chance to provide correct images in the service of justice. There has to be a desire to do the right thing and abolish from Standard Operating Procedures the pressure to cram as many images as possible on to a memory card in the hope of cost-saving. Perhaps now is the time to revisit the ACPO digital imaging handling guidelines, for the first time in nine years.

A JPEG workflow can never be a viable option to raw, no matter how good a system is, because of the blind spot in metadata tracking between the camera and copying images to a secure server. No system is tamper-proof, but the likelihood of someone being able to fabricate the provenance of a raw file is remote, as the computer scientists who might accomplish this would not have unsupervised access to the data. Conspiracy theorists take note.

A workflow should include raw data copied verbatim to a secure server. Any imaging software used should have the capabilities of Photoshopʼs History Log. The workflow should create two files; the secure original raw data and a working copy of that file. All changes made to the working copy could then be tracked and, if necessary, replicated in court to prove of its veracity.

Across the UK, the police take thousands of photographs, and archiving them all in raw requires a lot more server space than JPEG. Perhaps not all images need to be archived as raw, but the cost of extra server space does not compare to the costs of retrials, false convictions or letting the guilty go free.