Facial Mapping Evidence

Facial mapping by David Wells

Solicitor David Wells maintains that the reliability of facial mapping as evidence is very much open to question

Facial mapping evidence has been used in criminal trials in the UK for approximately fifteen years. An increase in its use has been noted following an increase in society’s reliance upon CCTV and the evidence it can provide.

Facial mapping evidence enables identification of an individual from a variety of imaging sources. The procedure involves taking two faces from a matching viewpoint, with similar lighting conditions, of a similar age, which are then aligned and scaled correctly. Facial geometry will only agree when these two faces are of the same person.

However, the reliability of facial mapping as evidence is open to question. ‘Expert’ testimony is often given in cases involving poor quality CCTV, and this in addition to other unreliable evidence may still result in conviction. In particular, poor lighting and low image quality inherent in CCTV systems means that the results from facial mapping are often low in confidence or inconclusive. Some research has revealed that there is often an insufficient awareness of the dangers of identification from such images.

CCTV is undoubtedly a valuable tool in our justice system, and despite some judicial concern of the validity of facial mapping evidence, the Court of Appeal have acknowledged, rather dangerously in my view, that such evidence can form the sole basis for a conviction. Although facial mapping evidence will on occasions prove to be of use, especially as it is open to the defence in any case to call its own expert in this field to rebut similar prosecution evidence, academics have argued that facial mapping is insufficient to support a conviction and some argue that it is inherently unreliable as a form of evidence.

The research available questions numerous aspects of the validity of facial image comparison evidence. Those concerns include that it is not regulated by a professional body, there is no required qualification of the system, there is no developed or agreed methodology (resulting in inconsistencies in the courts), that there are no established guidelines as to working practice and that there is no national or international database of facial features. It could reasonably be argued therefore that the current system of facial mapping appears to be significantly flawed and in need of subsequent reform in order to regain consistency throughout the UK courts.

Another issue with the validity of such evidence is that disclosed copies of CCTV footage will most likely be edited highlights prepared by the prosecution to suit the prosecution case. This often results in material important to the Defendant’s case being left out, and therefore does not give a complete picture to the jury. Furthermore, various concerns have arisen surrounding the ‘process’ of facial mapping, and in particular the fact that images are being compared, not faces. In addition, such evidence rarely provides conclusive evidence in favour of two persons being the same, in the way that fingerprinting or DNA evidence can.

Many mistakes in relation to facial mapping have been made in the past. Successful appeals have revealed that so called ‘positive’ mapping identifications introduced at trial have been shown to have been wrong, and where subsequent evidence has shown that a particular Defendant could not possibly have been the suspect. Like most forensic evidence suggestive of guilt, it always remains crucial for any Defendant not always to accept ‘expert’ evidence just because it comes from an ‘expert.’ It must always be a real consideration that such evidence is independently checked.

Other areas of concern have identified that CCTV footage is often of low quality due to a number of reasons including poor data compression methods, complicated processes for exporting video data which are generally specific to each manufacturer and systems are often set to maximize storage, which reduces the quality of the stored video data. Although image enhancement can uncover information within the image that is not initially evident, unfortunately CCTV images generally have low information content, so any improvement is only marginal at best. Some of the enhancement techniques which could be beneficial including adjusting the brightness and contrast levels of the video, applying sharpening or edge enhancing filters, increasing the colour saturation and removing noise. More sophisticated image enhancement systems are available, but again their effect is limited on low quality images.

Another issue with the reliability of facial mapping being used as evidence is that it is difficult to take any reliable measurements from images due to the visual distortion which is produced by the equipment and of different angles at which bodies or faces are photographed. Expert witnesses must rely on descriptive assessments, which can be easily influenced by prior knowledge of suspects. Observation of an actual live person provides more anatomical detail than a two dimensional small-sized image. The reliability of such evidence will depend on a variety of factors such as the nature and quality of the original images, the type of anatomical characteristics of a person of interest that can be reconstructed from images, the nature and quality of images of a ‘suspect’ or observation of a live suspect and knowledge and experience of the anatomist. Some crime scene images, despite being of good quality, show persons of interest only from the back, or depict only a portion of the face; and others, although showing numerous complete views of the entire person, are distorted by lenses resulting in a fuzzy image due to the equipment quality that it is difficult to form a good anatomical description.

In order to make the expert evidence more reliable it would be necessary to minimise possible bias and this could be achieved by ensuring that the expert is given only information relevant to the expert’s role in the case, because any extraneous information may produce a biased judgment. Another way in which such evidence could be made more reliable is by assessing the reliability of the expert’s statements and opinions through a peer review process, ideally by calling a second independent expert in each case; practically, by requiring experts to subject their opinions to the scrutiny of colleagues before they are presented to the parties and the court.

As with any activity involving human opinion, it is impossible to achieve absolute objectivity of results of an anatomical identification. Recent high profile cases which are still to be considered by the courts will undoubtedly raise some very interesting and challenging issues on the subject of anatomical identification and most likely ‘facial’ mapping. However, from a defence point of view the starting point must be that this area is by no means an exact science and therefore requires scrutiny like all other prosecution evidence.

* David Wells is a senior partner at Wells Burcombe LLP