Local Contrast Phase Descriptor for Quality Assessment of Fingerprint Images

Abstract

Fingerprint image quality is one of the main factors affecting the recognition performance of Automatic Fingerprint Identification System (AFIS). Therefore, analysis of fingerprint image quality is an important task during the acquisition. In this work, local contrast phase descriptor (LCPD) is used to analyze the texture quality of fingerprint images. Spatial and transform domain features computed using LCPD are fed to Support Vector Machine (SVM) classifier for fingerprint texture classification in wet, dry, and good class. Experimental evaluations performed on low-quality FVC 2004 DB1 dataset outperforms the current state-of-the-art methods. Therefore, utilizing the proposed method for quality control of fingerprint images during acquisition can help in improving the performance of fingerprint recognition system.