基于Contourlet与子空间投影的ICA手部特征多模融合识别算法

BASED ON CONTOURLET TRANSFORMATION AND ICA,REFERS TO THE PALM HORIZONTAL GRAIN MULTI- MODE FUSION RECOGNITION ALGORITHM

本文提出了1种基于Contourlet与ICA相结合的手部生物特征多模态融合的识别算法。这种多模融合的算法主要利用掌纹和指横纹特征2种模态进行融合,首先把掌纹和指横纹图像分别用Contourlet小波变换,得出一个低频分量和一些列的高频子带,其低频分量可体现纹理的概貌,高频子带可以概括纹理细节,将用Contourlet小波变换提取出来的掌纹与横纹的低频分量再用ICA算法对数据进行降维处理,再用归一化和简单加权融合的方法进行融合处理,最后用分类比对法计算出识别率。通过实验测试表明,两种模态融合的效果要优于单一的掌纹或指横纹的识别效果,证明了该算法的有效性。

This paper presents a combination of palm based contourlet and ICA multimode fusion reconition algorithm. The algorithm using mainly the palm and knuckleprint section. Firstly it transforms the sourse images of plamprint and knuckleprint by contourlet wavelet. After the inage transformed by contourlet,the low- frequency can reflect the texture profile,high frequency directional subband can reflect the contour,texture and so on. So the extracted can conduct dimensionality reduction for data by ICA algorithm.Then it implemonts the normalized and simple weighted fusion. Finally,the classification than obtained the recognition rate. Experimental results show that the fusion system recognition performance is superior to the performance of a single plam print recognition or knuckleprint,thus proving the validity of the algorithm.