Skip to Main Content
Based on gray scale vertical integral projection and wavelet decomposition, this paper presents a novel algorithm for hand vein feature extraction and recognition. Firstly, we design a hand vein image acquisition device, and Retinex method is used to enhance the image. Secondly, we analyze the performance of the four wavelet decomposition sub-band images and think that the low frequency sub-band image is the most suitable for hand vein features. We also think the image gray scale vertical integral projection is another very useful technique for hand vein feature extraction. Thirdly, the PCA transform matrix is built to reduce the dimension of the hand vein feature vector. Finally, we design a classifier based on SVM, and the experimental results demonstrate the high efficiency of the proposed algorithm in runtime and correct recognition rate.