Expertise in image analysis & comparison
Source level opinions in forensic comparison
In the varied fields of forensic comparison, trained experts compare two or more items to assist the court in determining if the items share a common source.
In forensic image analysis and comparison these types of comparison include whether a CCTV image and custody mugshot depict the same person, or if an item of clothing depicted in a photograph is of the same design and manufacture as a garment seized from a suspect.
In order to provide an evaluative opinion on the source of an item, the expert must be able to both observe reliable similarities and dissimilarities between the items under examination, and, understand the typicality or rarity of the observations within a relevant population. The expert will then convey these observations of similarity, dissimilarity and typicality as a strength of support, either for the hypothesis that the items share a common source or the hypothesis that the items are from different sources.
For example, a facial examiner provides opinion evidence on whether or not two images depict the same person by understanding the prevalence and commonality of facial features observed in the images, allowing them to evaluate the weight of any observed similarities or dissimilarities between two marks.
A problem arises in image analysis where experts are asked to provide source-level opinions and the expert does not possess demonstrable expertise in the prevalence of the types of items or features depicted in the images.
The cognitive nature of forensic comparison expertise
Within the field of cognitive psychology this understanding of prevalence and commonality is referred to as “statistical learning”. Research has demonstrated that forensic experts are better than novices at predicting the frequency of occurrence for certain features within their domain of expertise, including the fields of handwriting analysis and fingerprint comparison (Growns & Martire, 2020).
Statistical learning is derived from the extent of the expert’s knowledge, training and experience, not from the expert knowing in detail every single fingerprint or handwriting style in existence. Importantly, statistical learning is domain and stimuli specific, meaning that expert statistical learning in one type of object or subject matter does not transfer to other types. Therefore, having professed expertise in one domain (e.g. fingerprint comparison) is largely irrelevant to another domain of comparison expertise (e.g. facial comparison, see Phillips et al. 2018).
Critically, the expertise of forensic experts must demonstrable (i.e. an expert should be able to ‘show it’ see Martire et al. 2025). Within the field of psychology, expertise is broadly defined as being able to consistently demonstrate superior performance on a specific task in relation to novices (Ericsson & Lehmann, 1996). For example, an individual with commensurate knowledge, experience and training in the manufacture and distribution of firearms could provide an expert opinion as to the identification or comparison of a gun depicted imagery. However, simply having knowledge and experience in a subject does not necessarily equate to expertise. That individual would still have to be able to demonstrate that they can do the task beyond the capabilities of a lay person (Towler et al., 2018).
Superior performance at a task can be demonstrated by participation in formalised proficiency testing and validation studies, as well as published ‘black box’ error-rate studies, as has been done for fingerprint comparison (Ulery et al., 2011), facial comparison (Phillips et al., 2018) and vehicle identification (Bruehs et al., 2022).
Image analysis and subject-matter expertise
If an image analyst cannot demonstrate specialist knowledge, experience or training in a type of item depicted in imagery it is questionable whether that expert is any better than a lay person it determining how typical or rare any observed similarities or dissimilarities are.
This does not, however, preclude image analysts working with other forensic practitioners with demonstrable expertise in the relevant subject matter.
Take, for example, the comparison of items of clothing depicted in CCTV. A demonstrably competent expert in forensic image analysis may be best placed to assess the quality of the CCTV and the reliability of any observed similarities or dissimilarities within the constraints of the image. A second expert could then provide opinion on typicality of the observations derived from expertise in the design, manufacture and distribution of clothing. Butchart & Harrison, (2022) described this niche field as forensic garment analysis, and contrast the specialism to the informal clothing analysis that has typically been carried out by police investigators, as well as forensic image analysts (e.g. “clothing identification frequently limited to internet searches” where the individual is “unable to present expert interpretive evidence”).
Forensic science regulation and expertise
Clause 98.2.11 of the FSR CoP provides specific requirements for the provision of opinion evidence in image comparison, including the prerequisite for expertise in both imagery and the content of the image (i.e. the subject matter):
“Any practitioner proposing to give opinion evidence shall be an expert in all relevant aspects they intend to give an opinion on. Expertise in CCTV, video, imaging, enhancement etc. does not equate to expertise on the content of the image. Practitioners may highlight features of note such as logos or damage features; however, unless they are also an expert in the content of the images, practitioners should not attempt to give opinion evidence on the meaning of a comparison between the objects in question.”
Clause 98.2.11 further clarifies that:
“Image analysis requires specific subject matter expertise of both the system and the subject to be analysed.”
The basis for these requirements are derived from a 2019 notice issued by the Regulator (Forensic Science Regulator, 2019), which was prompted by the following issues:
“The Regulator has been made aware of incidents where image analysis experts appear to have failed to stay within the bounds of their expertise and non-experts appear to have provided opinion evidence on footage which they were not qualified to do.
Imagery experts have also presented to the Regulator a number of errors caused by comparison experts failing to understand the limitations of imagery and artefacts that may be present and/or failing to communicate effectively to the courts the limitations of work carried out.”
This notice details four principles that must apply when expert opinion relating to image comparison is presented. These principles have a basis in the legal obligations of an expert witness in England and Wales, e.g. that the expert provides opinions within their area of expertise (Criminal Procedure Rules, 2020) and that the expert is competent to provide the opinion (Criminal Practice Directions, 2023).
In relation to image comparison principle 2 of the FSR Notice is of particular importance:
“Principle 2: The person proposing to give opinion evidence must be an expert in all relevant aspects they intend to give an opinion on.
a. The proposed expert must be an expert in the subject matter on which opinion is expressed.
b. Expertise in CCTV, video, imaging, enhancement etc. does not equate to expertise on the content of the image.
c. Unless they are also an expert in the content of the images, imagery experts must not attempt to give expert opinion evidence on the meaning of a comparison between the objects in question.
d. An expert in a class of comparison (e.g. facial comparison) or in a specific field (e.g. Nike training shoes or Ford cars) but who does not have expertise in imagery must ensure that, in any case where the quality or nature of the image may affect the interpretation, an imagery expert has assessed the imagery in question to identify any artefacts or issues that could affect the comparison.”
If an image analyst does not have demonstrable subject matter expertise in the items or features depicted in the images, the analyst is not, within the context of the FSR CoP, qualified to give an expert opinion on these subject matters (clause c.).
Within the confines of the FSR CoP, if able to demonstrate expertise in imagery, the analyst at most “may highlight features of note such as logos or damage features” (clause 98.2.11 of the FSR CoP).
Conclusions
It is not uncommon for experts in image analysis to provide opinion on the comparison of objects outside of their domain of expertise (Hak, 2019). However, in no other field of forensic comparison does an expert routinely provide an opinion on subject matters that they are not an expert in. For example, a fingerprint expert will not provide opinion on a shoe mark. By providing opinions outside of the domain of expertise there is a highly significant risk that image analysis experts do not sufficiently understand the prevalence and typicality of the items and features that have been observed. As a result, their opinion will only consider the similarity of two observed features or items and they cannot provide a properly balanced evaluative opinion (Tully & Stockdale, 2019).
References
Bruehs, W. E., Tucker, N. M., & Meline, K. A. (2022). Observer determination of the make, model, and year of questioned vehicles. Journal of Forensic Sciences, 67(1), 200–206. https://doi.org/10.1111/1556-4029.14944
Butchart, A., & Harrison, K. (2022). Forensic Garment Analysis. In Encyclopedia of Forensic Sciences: Volume 1-4, Third Edition (Vol. 2). Elsevier Ltd. https://doi.org/10.1016/B978-0-12-823677-2.00072-6
Ericsson, K. A., & Lehmann, A. C. (1996). EXPERT AND EXCEPTIONAL PERFORMANCE: Evidence of Maximal Adaptation to Task Constraints. Annual Review of Psychology, 47(1), 273–305. https://doi.org/10.1146/annurev.psych.47.1.273
Growns, B., & Martire, K. A. (2020). Human factors in forensic science: The cognitive mechanisms that underlie forensic feature-comparison expertise. Forensic Science International: Synergy, 2, 148–153. https://doi.org/10.1016/j.fsisyn.2020.05.001
Hak, J. W. (2019). Evaluation of the Forensic Science Regulator’s recommendations regarding image comparison evidence. Forensic Science International: Synergy, 1, 294–297. https://doi.org/10.1016/j.fsisyn.2019.07.005
Martire, K. A., Neal, T. M. S., Gobet, F., Chin, J. M., Berenguet, J. K., Edmond, G. (2025). Psychological insights for judging expertise and implications for adversarial legal contexts. Nature Review Psychology, 4, 264-276. https://doi.org/10.1038/s44159-025-00430-4
Phillips, P. J., Yates, A. N., Hu, Y., Hahn, C. A., Noyes, E., Jackson, K., Cavazos, J. G., Jeckeln, G., Ranjan, R., Sankaranarayanan, S., Chen, J.-C., Castillo, C. D., Chellappa, R., White, D., & O’Toole, A. J. (2018). Face recognition accuracy of forensic examiners, superrecognizers, and face recognition algorithms. Proceedings of the National Academy of Sciences, 201721355. https://doi.org/10.1073/pnas.1721355115
Towler, A., White, D., Ballantyne, K., Searston, R. A., Martire, K. A., & Kemp, R. I. (2018). Are Forensic Scientists Experts? Journal of Applied Research in Memory and Cognition, 7(2), 199–208. https://doi.org/10.1016/j.jarmac.2018.03.010
Tully, G., & Stockdale, M. (2019). Commentary on: Hak. Evaluation of the Forensic Science Regulator’s recommendations regarding image comparison evidence. Forensic science international: Synergy 2019; 1(1). Forensic Science International: Synergy, 1, 298–301. https://doi.org/10.1016/j.fsisyn.2019.09.006
Ulery, B. T., Hicklin, R. A., Buscaglia, J., & Roberts, M. A. (2011). Accuracy and reliability of forensic latent fingerprint decisions. Proceedings of the National Academy of Sciences of the United States of America, 108(19), 7733–7738.