基于掌纹图像相关匹配的重复定位技术

Repetitive Positioning Technology Based on Correlation Matching of Palmprint Image

基于每次测量时所采集掌纹图像之间同时存在平移和旋转变化,先对图像进行空域旋转校正,然后利用相位相关方法进行相关匹配,并通过二次曲面拟合方法使平移变化的检测精度达到亚像元级,从而达到精确定位.该方法已成功应用于掌纹图像匹配定位中,对于无噪声干扰的掌纹图像,平移检测精度达到0.01像元,旋转检测精度达到0.1°.

Because of the concurrence of translation and rotation between the different palmprint images, the algorithm presented is based upon phase correlation with rotational compensation, and then paraboloid surface fit will be conducted in order to improve matching accuracy to subpixel level. The algorithm proposed has been successfully applied to matching and locating palmprint images. And for the noisefree palmprint images, The matching accuracy is more than 1/100pixel for translation and 1/10degree for rotation.