摘要
以掌纹为研究对象,提出基于分块2DPCA和二次2DPCA相融合的特征提取算法,并选择RBF神经网络作为系统分类器。首先对提取的ROI区域进行巴特沃斯-小波去噪,然后通过改进的算法提取得到最终的特征向量,最后由RBF神经网络进行分类识别。通过Poly-U掌纹库的仿真实验表明,此方法具有可行性和有效性,识别时间明显缩短,且保持了较高的识别率。
Taking the pahnprint as the research object, in this paper we propose a feature extraction algorithm which uses the integration of modular 2DPCA and the second-order 2DPCA, and select RBF neutral network as the classifier of the system. First, Butterworth-wavelet denoising is imposed on the ROI region extracted, and then the ultimate feature vector is obtained through the extraction with the improved algorithm, at last the RBF neutral network is employed for classification and recognition. This method is demonstrated by the simulation experiment on Poly-U palmprint library to be feasibility and validity, it clearly costs less computational time while keeps pretty high recognition rate.
出处
《计算机应用与软件》
CSCD
北大核心
2013年第6期269-271,280,共4页
Computer Applications and Software
关键词
掌纹识别
巴特沃斯-小波
分块矩阵
RBF神经网络
Palmprint recognition
Butterworth-wavelet
Partitioned matrix
RBF neutral network