Aspects of Appearance - Eyes, Hair, Skin, Race, Ethnicity

Original author and date: Glenn Pearson, April 29, 2009

The following tables show the “old code” and “old string” values that are the default for a 0.6.x install, and the “new “replacement values chosen for LPF, based on the standard mentioned beneath each table. Also indicated are speculative possible mappings if LPF needed to interchange data with a default Sahana instance.

Field code: opt_eye_color

Source for new code & strings: FBI’s National Crime Information Center (NCIC) 2000 – Personal Descriptors. (For a full reference to NCIC 2000, see the CASE STUDY page concerning Location Hierarchy.)

Field code: opt_skin_color

Source for new code & strings: FBI’s National Crime Information Center (NCIC) 2000 – Personal Descriptors

COMMENTARY: This representation does not make much sense if considered as mapping to absolute skin tones, but as a collection of words that people might say as a description. Terms like “light”, “medium”, “dark” seem to have nothing to qualify, and perhaps are most meaningful in conjunction with a “race” selection, i.e., dark[-skinned] white, light[-skinned] black. In Sahana usage, it is not expected that color matching against a absolute standard would be possible. Skin-tone systems that use color-chip matching [1][2][3] or instrumental readings [4][5] are thus not so useful. It is possible to infer skin tone from photographs if the image includes a calibrated color chart [6], or the photographic setup (camera, lighting, subject placement) is calibrated. Photography of driver's licensee's may approach the latter, but the accompanying calibration images needed for skin tone assessment would not be available to Sahana.

Skin Tone References:

[1] Felix von Luschan, “Volker, Rassen, Sprachen”. For a modern rendering of historic von Luschen color tiles, see C. Burnett, http://commons.wikimedia.org/wiki/File:Felix_von_Luscahn_Skin_Color_Chart.svg

[2] Jean de Rigal, Marie-Laurence Abella, Franck Giron, Laurence Caisey, Marc Andre Lefebvre, “Development and validation of a new Skin Color Chart”, Skin Res. & Tech., 13(1), Wiley & Sons, pp 101-107, 22 Jan 2007.

[3] Others mentioned in [4] include the Hyperpigmentation Scale of Dr. Susan Taylor, and the commercial IMS Human Skin Tone Chart.

[4] Gary Grove, Charles Zerweck, Jonn Damia, “Human Skin Coloration using the RGB Color Space Model”, 4th Intl Symp, L'Oreal Inst for Ethnic Hair & Skin Res, Nov. 9-11, 2007. This reports a cyberDERM/Cortex joint project to develop a low-cost skin tone sensor.

[5] As reported in [4], Cortex Technology's DermaSpec and Minolta Chromameter are traditional, expensive instruments for skin tone evaluation.

[6] J. Marguier, N. Bhatti, H. Baker, M. Harville, S. Su¨sstrunk, “Assessing Human Skin Color from Uncalibrated Images”, Int J Imaging Syst Technol, 17, 143–151, 2007, Wiley; Online (www.interscience.wiley.com). DOI 10.1002/ima.20114 . also: http://infoscience.epfl.ch/record/104380/IJIST_ICIP_07_marguier.pdf

Field code: opt_hair_color

Sources for new code & strings:

• FBI’s National Crime Information Center (NCIC) 2000 – Personal Descriptors
• Interstate Identification Index (III)

Field code: opt_race

It is proposed to use the US federal designators for Race (and possibly Ethnicity), in spite of their manifest shortcomings, because these are widely used in America for both investigative missing person reporting and health research analysis. (Alternatively, and friendlier, one could just have a free-text field to let the person reporting a missing person use his/her own words.)

In addition, the display_order field is set, giving:

Sources for new code & strings:

Center for Disease Control (CDC) Codes for Race and Ethnicity, circa 2000. The proposed categorization here reflects a minimal top-level hierarchy imposed government wide by the Office of Management and Budget (OMB). “New Code” values are the CDC “Hierarchical Codes”. Also shown is NEMSIS Encoding (Race), which has same hierarchy but different code values. CDC has defined additional sub-categorization codes (not shown), developed in conjunction with the Bureau of Census. Note: Census in 2000 allowed multiple races to be selected by an individual. NCIC 2000 and Sahana allow a single choice.

Possible Future Field code: opt_ethnicity [possibly as a checkbox]

Sources for new code & strings:

Center for Disease Control (CDC) Codes for Race and Ethnicity, circa 2000. The proposed categorization here reflects a minimal top-level hierarchy imposed government wide by the Office of Management and Budget (OMB). “New Code” values are the CDC “Hierarchical Codes”. Also shown is NEMSIS Encoding (Ethnicity), which has same hierarchy but different code values. CDC has defined additional E1 sub-categorization codes (not shown), developed in conjunction with the Bureau of Census. Note: Census in 2000 allowed multiple races/ethnicities to be selected by an individual.