• P-ISSN 0974-6846 E-ISSN 0974-5645

Indian Journal of Science and Technology


Indian Journal of Science and Technology

Year: 2016, Volume: 9, Issue: 45, Pages: 1-11

Original Article

Latent Fingerprint Recognition using Hybridization Approach of Partial Differential Equation and Exemplar Inpainting


Objectives: Biometric based Fingerprint recognition system is one of the well-adapted approaches in online security prospects. However, due to user-friendly behaviour of advanced computing systems, biometric approach is not much in use now days. The approach left its footprints only with the applications to identify criminal activities at crime scenes where fingerprints are mainly available in latent form. Latent fingerprints are the accidently left finger skin impressions by criminals. These impressions are invisible for the naked human eye and usually captured with lasers, chemical, powders etc. These captured latent fingerprints carries less minutiae information with distorted ridges and high level of pattern overlapping. So, it is not easy to identify the criminals with partial fingerprint information. Methods/Statistical Analysis: In this paper, hybrid approach of Exemplar In painting and Partial Differential Equation is used to fill up the distorted ridges. The main goal of this work is to present the framework to reconstruct the latent distorted fingerprint and further use them to find the best match for those enhanced reconstructed latent fingerprints. For the experimentation of proposed hybrid concept, IIIT Delhi latent and NIST SD-27 databases of fingerprint are used. In this experimentation, different fingerprint enhancement filters like Canny Edge Detection Filter, Prewitt filter, Laplacian Filter, Sobel Filter, Gaussian Low Pass Filter, Gaussian High Pass Filter are used. Findings: Different filters show different performance of dataset images for latent fingerprint recognition. The overall performance of the automated latent fingerprint identification approach is analysed in terms of false acceptance rate and genuine acceptance rate. Application/Improvements: In this way, latent fingerprint can be used for the recognition of criminal activities at crime scenes where fingerprints are mainly available in latent form. The overall concept shows better results for canny filter as compare to other considered filters in enhancement.

Keywords: Binarization Approach, Criminal Activities, Exemplar Inpainting, Latent Fingerprint, Minutiae Extraction, Partial Differential Equation


Subscribe now for latest articles and news.