梯度法在大景深非接触掌纹识别中的应用

Application of Gradient Method in Large Depth of Focus and Non-Contact Palmprint Recognition

提高非接触掌纹识别性能是当前掌纹识别领域的研究热点。针对大景深非接触掌纹图像特点,应用解决缩放与旋转等现象的预处理方法 ,同时在特征提取上考虑了掌纹纹理的方向性,采用小波分解下的梯度法对ROI进行处理,获得一系列数据作为图像特征以用于识别。运用自建的SUT手形图像库,对本文算法进行了实际测试。结果表明:系统的正确识别率达到90%,优于一些经典的子空间法,提高了大景深非接触掌纹识别系统的性能。

At present, it is a research hotspot in palmprint recognition feld to improve the performance of non-contact palmprint recognition. The preprocessing method is used to solve the phenomena about scale, rotation and so on according to the characteristics about large depth of focus and non-contact palmprint images. At the same time, with considering the directions of palmprint texture on feature extraction the gradient method under wavelet decomposition is adopted to deal with ROI. A series of data is obtained as image features for recognition. The method is tested on the basis of self-built SUT hand image database. The results show that the correct recognition rate for the system is 90% and it is superior to some classical subspace methods. It raises the performance of large depth of focus and non-contact palmprint recognition system.