Improved Method Of Latent Fingerprint Recognition Using Descriptor-Based Hough Transform And Texture Features
Fingerprints lifted from crime scenes is a routine procedure used for identifying suspects that is extremely
important to law enforcement agencies and forensics. Latents are partial fingerprints with small area and containing large
distortion. The noise characteristic and small number of minutiae makes the latents extremely difficult to automatically
match the fingerprints that are stored in law enforcement databases. The proposed approach provides an efficient algorithm
for identification of individual latent fingerprints. Three texture based features are used to improve the matching accuracy.
These features include 1. Entropy coefficient 2. Correlation coefficient 3. Energy coefficient. The proposed approach
provides better results as they are used to match features of fingerprint images.