NCJ Number
203114
Journal
Journal of Forensic Sciences Volume: 48 Issue: 5 Dated: September 2003 Pages: 973-984
Date Published
September 2003
Length
12 pages
Annotation
This paper proposes a novel digital image restoration technique for restoring defective (lacking information needed for completeness) or contaminated (undesirable information included) fingerprint patterns, based on the AM-FM (amplitude modulation-frequency modulation) reaction-diffusion method.
Abstract
The method involves computing a complex-valued extension of a processed fingerprint image with Hilbert Transform and estimating the parameters of the AM-FM functions demodulated from the complex-valued extension. In the reaction-diffusion process, the estimated parameters are used to reconstruct fingerprint patterns. The method showed its potential application to fingerprint-pattern enhancement in the recognition process but not in the identifying process. The method can be used to improve the legibility of the input image for automated systems such as the automated Fingerprint Identification System (AFIS). The improvements achieved comply with the guidelines provided by the Scientific Working Group on Friction Ridge Analysis, Study and Technology, the Scientific Working Group on Imaging Technology, and an AFIS working group within the U.S Justice Department's National Institute of Justice. Since the method described in this paper produces fingerprint patterns that correspond to "oriented texture features," it cannot create any specific patterns that are different from those features. The authors caution that when applying the proposed method to fingerprint pattern restoration, the size of the poor clarity region cannot be of an order greater than that of the fingerprint ridges; otherwise "false alarm features" may be generated. 15 figures and 26 references