基于曲波变换决策融合的掌纹识别

Palmprint Recognition with Decision-making in Curvelet Domino

用曲波变换提取出掌纹图像在不同尺度下的图像特征,通过降维处理的主成分分析(PCA)方法,由径向基函数网络训练或判决,按照用户的要求,选择不同的决策规则,对各种尺度下的识别结果进行融合,可达到较高的掌纹识别率。

Curvelet transform is a multi- scale method that can represent curves most sparsely. The main feature of palmprint images is that it is made up of several curves. This paper uses the curvelet transform to get the features information of palmprint images under different scales,deals with the information by dimension reduction of PCA（ Principal Component Analysis）,then provides the information for RBF network to study and make decisions. Finally,according to the customers＇ requirements,this paper chooses decision rule to get the fusion of the results and attain the goal of palmprint recognition. It has been proved by the experiments that this method can fully make use of the features information of images and can get higher recognition rate than other methods.