The T-Model initially simplified the Osterburg grid of 8,591 cells into a 23,136 mini-cell grid structure based on the measurement of the average size ridge unit in order to make first "best guesses" for what are the discriminating values for fingerprint ridge features.  Each Osterburg 1mm x 1mm cell was divided into five mini-cells each with an area of approximately .45mm x .45mm.  The Osterburg cell distributions, frequency of occurrence and discriminating values weights were then re-defined whereby combinative ridge formation types are redistributed and placed back into original form within a “mini cell” framework.

Given that the average fingerprint impression contains relatively equal amounts of ridge and furrow detail, the Osterburg “empty cell”, which represents approximately a 2mm length of continuous ridge detail, was restructured to contain five .45mm x .45mm mini cells each bearing half unrecorded furrow detail and half continuous ridge detail.  The  frequency of occurrence results for individual and compound ridge formation types found in Osterburg’s cell structure were redistributed based on the number of specific ridge unit types found in each mini-cell.

For example, an Osterburg 1mm x 1mm cell containing a single “spur” is restructured as five ridge unit types in five mini-cells as follows:  1 ending ridge unit, 1 bifurcating ridge unit, 1.5 continuous ridge units and 1.5 furrow units.  As a result of this redistribution, approximately 19,711 furrow units were found to exist in the total grid of 42,955 mini-cells.   Since ridge and furrow units share a common border (which means ridge and furrow edge contours equal) and since furrow width is impacted by ridge edge contour, the furrow unit is omitted from the T-Model.  The total number of mini cells used to define cell distribution and frequency of occurrence changes from 42,955 to 23,136.  As a result, new cell distributions with new frequency of occurrences based on individual and independent ridge unit types were then defined. 

The resulting discriminating values were then tested using "close match" experiments to determined whether or not the predicted values were consistent with nature, i.e., observed values from experiment).  Based on results from these experiments (see Validation Studies) the discrimnating values for most fingerprint rdige features were refined, some significiantly so. 

Current "best guess" conservative discriminating values (match probabilities) for fingerprint ridge features are as follows: 

NEXT PAGE >>> 

Discriminating Value for Ridge Unit Types

The discriminating value for ridge unit types is defined as a positive number based on the inverse or reciprocal of its frequency of occurrence.  Discriminating values for the most frequently occurring individual ridge formation events were defined based on frequency of occurrence in the 23,136 mini-cell framework (based on extrapolation of the Osterburg study).

Although based on empirical data, the continuous ridge, ending ridge, bifurcation and dot are the most frequent ridge unit types used by latent fingerprint examiners to establish inference for  identification, for purposes of a more complete model the frequency of occurrence for other, more rare types, were included, i.e. cores, scars, creases and pores.

Note:  The discriminating value for ridge unit types must also consider the frequency of ridge unit position or intervening ridge count to its nearest neighbor (see Intervening Ridge Count).  The product of these frequencies define a ridge units total discriminating value (absent qualitative reduction factors).