Neural Network Based Fingerprint Classification using DCT Compression

Abstract

Recognition of the people by the method of their characteristics of biometric is very much popular and common in the society. But among all this, fingerprint recognition is very vital technology for identification which is personal because of its structure which is very unique. Enormous quantity of fingerprint are gathered and then stored in the applications everyday in broad range. Image compression techniques are used in handling a large database of fingerprint images, for example, in access control and forensic science. Fingerprint images are used in various documents like gun registration, police records and passport which needs huge amount of storage memory. The various other fields of fingerprint image compression are log in authentication and other locks, library access and electronic configuration. It is used to validate transactions by requesting biometric authentication before orders are submitted or financial transactions are executed. In this novel approach is presented in which base matrix is constructed initially. Then it is divided into small blocks which are patches and after that quantization of coefficients are done and then coefficients are encoded by using methods of lossless coding. And at the last, features are extracted and are classified by using approach of neural network.

Keywords: Wave propagation, ray tracing, tri linear refractivity, ducting, anomalous propagation, PWE, HFSWR