Danish forensic procedure addresses subjectivity in face biometrics investigations
Forensic examinations of face biometrics for immigration cases in the Netherlands follow a procedure that both acknowledges and tries to mitigate subjectivity, which was explained to members of the European Association for Biometrics in an online presentation.
‘Forensic Facial Examination – transparency and uniformity in case work’ was presented by Dr. Trine Edvardsen of the Danish National ID Centre in a recent EAB Lunch Talk, based on a presentation earlier in the year at IAI in Omaha.
The Danish national ID Centre is an independent expert body which provides services to the country’s immigration authorities. Biometrics experts within the center focus on face and fingerprint modalities.
The center provides full morphological comparison of some cases, a more limited biometric screening procedure for others, and an assistance service for urgent cases. The latter can include advice on whether to escalate the given case to a full morphological comparison, Edvardsen explains.
Full morphological comparisons take a half-day, and are carried out with the biometric analysts to minimize the potential for bias. A double blind procedure with full documentation is carried out, with conclusions ranging from -3 to +3, with 0 meaning the analysis is inconclusive.
Edvardsen notes that in many countries, the term “conclusion” would be rejected in favor of “opinion,” but says that in the Danish language, “conclusion” is the preferred term, without necessarily imparting any different meaning from “opinion” in other situations.
She reviewed the procedural documents, which cover technical analysis, a facial feature checklist, side-by-side comparison and evaluation. The feature analysis is based on the FISWG guidelines.
Features are described at three levels, from basic outline to clearly resolved dimensions. Most images will be in between, at level 2 feature visibility.
The Centre began revamping its facial examination procedures just over two years ago, introducing a decision tool to increase the transparency of the complex decisions made.
In reviewing the elements of facial comparison that are considered vital to the final decision, Edvardsen notes the use of presence or absence of other distinct features as an added variable, “kind of like our wild card.” This considers distinguishing marks beyond the global features, such as a set of freckles or a visible scar.
Image suitability, comparability, feature visibility and similarity, and the presence or absence of other distinguishing figures thus all contribute to the final assessment by the Centre.
Edvardsen presented the Excel spreadsheet the procedure is documented in, and the weighting of the criteria. The model does not determine the final score, however, but rather the outer bounds of allowable judgements the case workers can make.
“We’re taking the feeling out of this case work,” Edvardsen says.
If less than 5 global features are visible, for example, judgements at the highest relative level of certainty are not available to examiners.
Efforts to further reduce the subjectivity in the system could include adjusting the image suitability criteria to draw on ISO standards for image quality.
Article Topics
biometrics | biometrics research | EAB | European Association for Biometrics | face biometrics | forensics | immigration | Netherlands
Comments