In this paper, the authors propose a latent-to-full palmprint matching system that is required for forensic analysis, and they introduce a robust algorithm to estimate ridge direction and frequency in palmprints is developed.
The evidential value of palmprints in forensics is clear as about 30 percent of the latents recovered from crime scenes are from palms. While palmprint-based personal authentication systems have been developed, they mostly deal with low resolution (about 100 ppi) palmprints and only perform full-to-full matching. The authors propose a latent-to-full palmprint matching system that is needed in forensics. Their system deals with palmprints captured at 500 ppi and uses minutiae as features. Latent palmprint matching is a challenging problem because latents lifted at crime scenes are of poor quality, cover small area of palms and have complex background. Other difficulties include the presence of many creases and a large number of minutiae in palmprints. A robust algorithm to estimate ridge direction and frequency in palmprints is developed. This facilitates minutiae extraction even in poor quality palmprints. A fixed-length minutia descriptor, MinutiaCode, is utilized to capture distinctive information around each minutia and an alignment-based matching algorithm is used to match palmprints. Two sets of partial palmprints (150 live-scan partial palmprints and 100 latents) are matched to a background database of 10,200 full palmprints to test the proposed system. Rank-1 recognition rates of 78.7 percent and 69 percent, respectively, were achieved for live-scan palmprints and latents. (Published Abstract Provided)