Fingerprint image classification by local area analysis

摘要

This paper describes a new fingerprint classification algorithm based on the analysis of the local area in which singular point might exist and presents methods to eliminate false singular points made by noise. The algorithm detects singular point (core and delta) candidates roughly from 16x16 block-size directional image and analyzes the near area of each candidates from 8x8 block-size directional image. In this local area analysis, false singular point made by noise is eliminated and the precise core point and the type and the orientation informaiton of core point is extracted for the classification step. Using this information, classification is performed. The algorithm was tested on 11796 images and classification acuracy of 91.7