二维EMD分解提高PCA掌纹识别率

摘要

For improving the recognition rate of principal component analysis( PCA) which often used in palmprint online recognition system, a new palmprint recognition method with PCA and bi_dimensional emperical mode de-composition(BEMD) is proposed in this article. An image can be decomposed with BEMD in frequency domain, so it can be processed in different frequency domains. Because the low frequency part of palmprints sis often influenced by the background, the high frequency information is used in our experiment to highlight the personal characristics, and as the result, the recognition rate is improved and the speed is faster. The result of experiments with the palm-print database of Hong Kong Polytechnic University shows the recognition rate of BEMD and PCA is more higher than traditional PCA, and the results also indicate that this method plays an important role in both theoretical re-search and practical application.%为了提高常用于在线掌纹识别的PCA方法的识别率，提出融合BEMD技术的PCA掌纹识别方法。二维EMD技术能够在频率域内实现图像的多层分解，在不同频段内对图像进行处理。掌纹图像的低频部分容易受到背景等因素的影响，所以实验中提取、利用掌纹高频信息，去除低频信息，充分利用掌纹中的个人特征信息，抑制干扰，提高识别率。基于香港理工大学掌纹数据库的仿真结果显示，这种方法的识别率远高于传统PCA方法，体现了一定的理论研究意义和实用价值。