Algorithm For Automatic Fingerprint Classification

Fingerprint classification provides indexing mechanism for a large fingerprint database. Fingerprint image is classified into predefined categories. It can greatly reduce matching time because the matching is implemented in a sub-category. This paper presents a classification algorithm that is used for police system. Our algorithm classifies fingerprint image based on number and position of singular points. Because the police fingerprint database is very large and the fingerprint image is often noisy, we paid more efforts on extraction of singular point from the noisy image and improve the fingerprint classification speed. We have obtained some result in the following research areas: 1) Fingerprint extraction. We extract fingerprint image by two steps: coarse extraction and fine extraction. 2) Singular point extraction. We proposed a singular point extraction algorithm based on multi-characters. It uses direction information, neighborhood information and human experience to find singular point. 3) Fingerprint classification. We proposed an improved fingerprint classification algorithm. It receives better result on loop fingerprint with ridge tracing. Experimental results show that our algorithm is capable of correctly and robustly extracting singular point, rapidly and efficiently performing fingerprint classification.