A Fast Fingerprint Classification Algorithm by Tracing Ridge-Flow Patterns

Abstract

We propose a novel algorithm to classify fingerprint images using the ridge flow patterns. In Fingerprint Verification Systems (FVSs), classification of the fingerprint images saves time and increases accuracy, as compared to the sequential search method. In this paper, we have first extracted the High Ridge Curvature (HRC) region using the Sobel operator and Gabor filter. Once the region is extracted, the ridges within the region are traced in both the directions, starting from any point on the ridge. Vectors are drawn at the end points of the ridge. The vectors drawn at the end points are able to determine the class to which a fingerprint image belongs. In case of a whorl, the co-ordinates of the ridges from the starting point are monitored in order to determine the class. The algorithm is very fast as it eliminates the core point detection, and is based on simple ridge-flow connectivity. The implementation was done on Matlab, and the overall accuracy of classification without rejection was 98.75%.