Forensic image comparison

Forensic image comparison · Overview

Overview

Morphological forensic image comparison concerns judicial and administrative proceedings in which a person shown in a questioned image is compared with a known comparison person. In practice this mainly concerns road-traffic regulatory offences, but also criminal matters, video surveillance, documentation of individual acts, or other constellations in which an image record is expected to support an evidential question. The focus is not a mere impression of similarity, but the expert examination of the supporting features and their limits. The starting point is therefore always the question whether the available material permits a reliable comparison at all and which feature areas can actually be used for that purpose.

The methodological core is simple and demanding at the same time: the visible face or other recognisable body region is broken down into individual features. These features are described, examined for recognisability and comparability, and then weighted in an overall expert assessment. Precisely because the procedure is not standardised, the reasoning in the opinion itself must be sustainable. A result without disclosed foundational facts is insufficient.

Forensic image comparison · Methodological framework

Methodological framework

Scientific identity assessment avoids precisely what is typical in everyday recognition: wholeness, speed, and the premature switching between “identical” and “not identical”. The AGIB standards instead require careful analysis of individual features, a written justification, and the possibility of graduated assessments of similarity. Preselection situations must be disclosed explicitly because they alter the later interpretation. Where a comparison person has already been named because of apparent similarity or selected from a larger candidate pool, this changes the interpretive framework of the later assessment: similarities must then be weighed more cautiously, whereas dissimilarities may gain particular importance. Likewise, partial opinions or mere rapid assessments are expressly not recommended as a rule.

In practice this means that the conclusion is not carried by a pointed assertion, but by a traceable chain of examination: which features are actually recognisable in the questioned image? Which of these are visible in the comparison images in a genuinely comparable way? Which differences can be explained technically, perspectivally, or by variability, and which are to be regarded as robust dissimilarities?

Only such features are evaluated as are actually recognisable and comparable on both image sides. Particular account must be taken of perspective, illumination, sharpness, compression, covering, facial expression, age-related change, and the comparability of the reference material. The subject of the opinion is not whether two images “look similar”, but whether the ascertainable features support an expert conclusion on identity or on its exclusion. The examination of image suitability and the actual expert assessment must therefore be kept methodologically distinct, even though both may be closely connected in the proceedings.

That facial features play the greatest role has long been established in the professional literature. At the same time, that literature emphasises that image quality and the number and character of the recognisable features are decisive for the strength of the conclusion. A good comparison therefore does not live from a single striking structure, but from an overall finding based on several usable features.

Forensic image comparison · Image material and comparability

Image material and comparability

Every identity question begins with an examination of the material. What matters are not only resolution or file size, but the interplay of sharpness, contrast, perspective, illumination, compression, covering, facial expression, and the number of recognisable features. More recent work on image suitability expressly shows that a mere reliance on pixel counts or a single threshold is too simplistic. Depending on the starting position, it may therefore first be necessary only to examine image suitability; in some cases prior technical visualisation is sensible.

For practice this leads to a sober consequence: a formally large image may be methodologically weak, whereas a limited measurement image may still permit a sustainable examination under favourable conditions. Conversely, a poor image must not be overinterpreted merely because the procedure or the procedural setting calls for a clear answer. Precisely here an expert examination parts company with a mere expectation of the result. Where the material does not support the evidential question, this must be stated openly — even where a person has already been named or preselected beforehand.

Comparability also presupposes suitable comparison images. Ideally, viewing direction, head position, and technical conditions correspond as far as possible. Where that is not the case, the limitation must be stated openly. If needed, technical visualisation or the targeted production of comparison images may be useful; but this is an aid and not the opinion itself.

Image processing in an expert opinion: permissible and impermissible

Technical adjustments may be required in order to improve comparability when image documents are examined. This includes in particular the alignment of size, crop, position, brightness, and contrast. Such measures serve only the visualisation and comparative examination of information already present in the image.

Impermissible, by contrast, are interventions that alter, add to, or retouch image content. The original file or original image always remains the reference of the opinion. What matters for the assessment is not a single technical parameter, but the number of features recognisable in the case, their recognisability, and their characteristic value. The opinion therefore follows a traceable methodology without concealing the limits of the material.

Forensic image comparison · Typical constellations

Typical constellations

Typical case groups are driver identification in regulatory offence proceedings, criminal matters with surveillance footage, documentary and passport cases, as well as constellations in which a named person has already come into focus through prior investigative steps.

For practice this means that not the label of the case but the combination of evidential question, material basis, and prior history is decisive. Two apparently similar image cases may therefore require methodologically quite different treatment.

Forensic image comparison · Conclusion and limits

Conclusion and limits

A forensic image comparison can support identity, argue against it, or reach a methodologically open result. It is therefore not a procedure that necessarily ends in confirmation. Precisely where the material is weak, non-comparability or a restricted evidential value may itself be the professionally correct result.

The expert conclusion is neither a black box nor a mathematically spurious precision. The literature has long emphasised that, for biostatistical exactness, the underlying data are lacking and that probability should therefore be formulated as a verbally reasoned probability statement with a traceable justification. It is likewise pointed out that the number, recognisability, and character of features must be weighed together instead of multiplying rare individual features schematically.

The opinion also regularly stands under reservations. These include, in particular, close kinship, marked age-related change, alteration of outward appearance, problematic comparison images, or other documented limitations of the material. Such reservations are not a weakness, but part of a clean method. They make visible on what the conclusion rests and where its limits lie. No automated facial recognition, no schematic pixel threshold, and no assertion without a sustainable basis of comparison can replace this traceable case-specific assessment.

Forensic image comparison · Initial enquiry

Initial enquiry

For an initial classification, a few but precise details are usually sufficient: what concrete evidential question is at issue? Which questioned image is available, and in what form? Are comparison images already available, or would comparison material first have to be created? Has a preselection already taken place, for example through witnesses, wanted-person measures, internal investigations, or technical hit lists?

The more clearly this initial position is described, the easier it is to determine an appropriate route of work: a preliminary review of image suitability, technical visualisation or conversion, or a fully reasoned comparison opinion.

Helpful for this are the best available questioned image, existing comparison images, information on exports or post-processing, and an indication whether a preselection, photo array, witness naming, or other prior identification has already taken place. This makes it possible to clarify quickly whether a preliminary review is sufficient or whether a full opinion is methodologically justified. Even the brief communication of the type of proceedings, the evidential question, and the state of the material considerably facilitates the first professional classification.

On procedure and the structured enquiry