基于MATLAB的SUSAN指纹细节点提取

SUSAN for extraction of fingerprint minutiae based on MATLAB

在指纹识别中,指纹细节点提取具有重要的意义,它直接关系到指纹匹配的可行性。指纹细节特征提取通常直接从灰度图像中提取特征点,虽然这种算法不必经过二值化和细化过程,但它要对处理后的指纹图象进行纹线修复,然后再实现细节特征提取。纹线修复往往比较耗时,步骤繁琐,同时对噪声比较敏感,甚至涉及到复杂的梯度和微分运算。本文基于简化运算的目的,首先建立USAN区域,再观察USAN区域的变化情况来判断特征点类型,最后提出了较完善的基于SUSAN原理的指纹特征提取算法,并且在MATLAB下进行仿真。实验结果证明了本算法对USAN的求和相当于求积分,计算简单、抗噪声能力强。

Minutiae extraction has great significance in fingerprint recognition, which is connected with the feasibility of fingerprint matching. The minutiae are usually extracted directly from the original gray-scale fingerprint images. The process of thinning and binary is not needed, but the algorithm should repair the ridge for the processed fingerprint images, and then extract. Sometimes, ridge repairing is time--consuming and tedious, and more sensitive to noise, or even related to complex gradient and differential operation. For the purpose of simplifying operations, USAN region is established firstly, then according to the changes of USAN regional to judge the feature points, finally a perfect SUSAN algorithm for extracting of fingerprint minutiae is offered, we also have a simulation based on MATLAB. Experimental results show that the sum for USAN equivalent to seek integral region in the algorithm, the operation is simple, and the algorithm has the ability of strong anti-noise.