指纹识别预处理算法的研究

Study on Preprocessing Algorithms of Fingerprint Recognition

实现一套主要包括指纹图像灰度归一化和均衡化、指纹图像分割、指纹图像二值化和二值化后处理及细化的指纹识别预处理算法。采用基于梯度的指纹分割方法,用连通域检测方法实现对分割后指纹图像的优化。在方向二值化的基础上引入"块"的概念,实现基于块进行的方向二值化。最后,以matlab7.0为开发平台对该方法进行验证,表明在同样的特征提取算法和匹配的条件下,改进的预处理算法识别率高于传统算法,所用预处理时间少于传统算法。

A set of fingerprint recognition pretreatment algorithms mainly including fingerprint gamma controller normalization and equalizing, fingerprint image division, fingerprint image binarization and refinement was established. The fingerprint dividable method based on trapezium was adopted, and the fingerprint image which had been divided by using connected examine ways was fined. Directional binaryzation by transfer of block technology was proposed. Moreover, a lot of experiments by the platform of matlab7.0 was made which proved that the new methods had the advantages of stronger recognition rate and quicker speed than tradition in the same condition for characteristic extraction and matching.