Skeletonization of binary images and finding of singular points for fingerprint recognition

The paper reviews a new modified Rosenfeld skeletonization algorithm of binary fingerprint images. Singular points are allocated on the basis of the skeleton. Skeletonization and allocation of singular points are the basic procedures for solving the problem of verification and fingerprint identification. The described algorithm analyzes the neighborhood of informative image points and performs actions based on analysis. A detailed analysis of neighborhoods facilitates to remove noises that are contained in the original image and can be recognized as singular points of the fingerprint. Advantages of the new method of skeletonization compared to the known described in this work are also examined. The method is implemented in C++. The procedures to visualize the final state of the skeleton are developed for monitoring the quality of its creation.