中文摘要：

提出了基于EMD和ICAI的掌纹识别方法，该方法利用二维EMD技术对掌纹图像进行分解，得到4个IMF分量和1个残余分量，用前4个IMF分量重构掌纹图像，然后利用ICA结构I提取EMD重构掌纹图像特征并进行识别。采用香港理工大学掌纹数据库进行实验，由于充分利用了掌纹图像的EMD高频细节特征。与ICAI、PCA方法相比，有较高的识别率和较快的识别速度，表明该方法有一定的理论研究价值和一定的实用性。

英文摘要：

A palmprint recognition method based on empirical mode decomposition （EMD） and architecture I independent components analysis （ICA I ） is proposed. Palmprint images are decomposed with the bidimensional EMD to get 4 IMFs and a remainder, then reconstruct the palmprint images with the forward 4 IMFs. When all processes above are finished, the next step is to extract ICA I feature from the reconstructed palmprint and recognise the reconstructed palmprint. The palmprint database of Hong Kong Polytechnic University is used in experiments. Due to many high frequency characteristic details in palmprint is employed, the recognition rate obtained is higher than ICA I and PCA, the speed of recognition is also faster. The results of the experiments convincingly indicate that this method will play an important role in both theoretical research and practical application.