随机选取ROI法在掌纹识别中的应用

Application of choosing ROI randomly in palmprint identification

从旋转后的掌纹图像中选取一块感兴趣区域进行识别是当前掌纹识别中常用的方法,但此方法会丢掉部分有用信息。针对此问题,在对掌纹图像建立坐标系的基础上随机选取多个感兴趣区域,用这些感兴趣区域来代表某幅掌纹图像,分别在这些感兴趣区域上用小波能量的掌纹识别算法进行识别。实验结果表明,使用此方法在掌纹识别的识别率上有较大改进。

Choosing only one region of interest from the palm-print image after rotating is a common method which is cur- rently used in palm-print identification, however it looses some useful information. To solve this problem, in this paper, the palm-print image is represented by some regions of interest, which are chosen from the palm-print image randomly, on the basis of creating coordinate system on it. Then the method of wavelet energy is applied on these regions of interest respec- tively. The experimental results verify that the random region of interest method improves recognition rate of palm-print i- dentification.