基于线性支持向量机的指纹图像分割方法*

针对现有指纹分割方法存在的不足，提出了一种基于线性支持向量机的指纹图像分割方法。该方法首先选择指纹图像子块的傅里叶频谱能量比和灰度对比度构成特征向量，利用线性支持向量机对这些特征向量进行分类，再运用二值图像形态学操作最终实现指纹图像分割。实验结果表明，与现有的一些典型方法相比，该方法对于指纹图像分割更为准确、可靠。

Fingerprint image segmentation method based on linear support vector machine

This paper proposed novel fingerprint image segmentation method based on linear support vector machine to overcome some shortage in the present literatures.The method firstly selected Fourier spectrum energy ratio and gray contract which were extracted from fingerprint image sub-blocks to construct feature vectors,and then classified these feature vectors by linear support vector machine,finally performed morphological operations to ultimately realize fingerprint image segmentation.Experimental results prove that this fingerprint image segmentation method is precise and reliable compared with other approaches.