Latent fingerprint recognition and categorization using Multiphase Watershed Segmentation

Latent fingerprints play a very important role in forensic applications to recognize the criminals. The latent fingerprints still face many issues due to the large amount of distortion. The sole purpose of this article is to improve the matching accuracy of latent fingerprints which are of bad quality. In this research work, we propose Multiphase Watershed Segmentation algorithm to refine the features collected from the poor quality fingerprint impression. Firstly all fingerprints of fine quality are acquired from the user dataset. Then we introduce noise models into image like Gaussian disturbance to test the matching process performance. The image is enhanced with the help of anisotropic filter. Weighted equalization of pores, line patterns and minutiae features are extracted and estimated. These are used to match from database to query using multiphase Watershed recognition algorithm. This algorithm is based on feature space calculation to discover and describe the primary feature in images. The features are vigorous to most of the image variations. The analyses are implemented in both frequency and spatial channels.