A NEW FINGERPRINT IMAGE ENHANCEMENT BASED ON DECIMATION-FREE DIRECTIONAL FILTER BANK

摘要

Fingerprint images are direction-oriented pattern formed by ridges and valleys. The dominant direction of ridge plays a very important role in the enhancement of fingerprints. This paper presents a new method of decimation-free directional filter banks (DFB)-based analysis for fingerprint enhancement. Its analysis outputs are shift-invariant directional images but not directional sub-band images in previous DFBs. The algorithm first decomposes a fingerprint image into eight directional images in the analysis stage; then obtains the orientation image based on directional energy distributions for each block and processes the directional images based on the orientation image in the processing stage; at last reconstructs them to the enhanced image in the synthesis stage. Experimental results show that the proposed method reduces the influence of noise, links the injured ridges and enhances the ridges' moving shape.