Hand Vein Recognition Technology Based On The Region Code

The hand vein recognition is a method of identification which is carried out byanalyzing the characteristic of the human body on the back of the hand vein. In thispaper, some of the key technologies of the hand vein recognition are researched, themain work is as follows:Firstly, the hand vein image acquisition. Image acquisition is the foundation asthe back of the hand vein recognition technology, First of all, according to thecharacteristics of the dorsal hand vein exists, the various components of theinstrument on the back of the hand vein harvesting comparison and strict selection of,and then the selected components formed into the hand vein acquisition instruments,In contrast to the results of various experiments, to obtain the placement of eachmember as well as an angle between the optimum physical parameters. Finally, theuse of adjusted the hand vein instruments image acquisition.Secondly, the global hand vein image preprocessing. After image gray scale andsize normalization, image segmentation, image enhancement and binarization,extracting the global image of the back of the hand vein network by the morphologyof set theory, and then further image denoising, smoothing, thinning and go "glitch".Then get the original back of the hand vein to a single pixel width of the back of thehand vein skeleton image.Thirdly, the dorsal hand vein skeleton image feature extraction and featurematching. Feature point extraction of hand vein image after pretreatment, the HandVein intersection of the obtained, based on the characteristics of the area codeextraction methods. Through hand vein image in the number of the endpoints, thenumber of cross-point, the slope of the cross point, the intersection point area code,and a code of each region crosspoint number of these five feature values to performthe dorsal vein image matching. Finally, sample library identification experimentsreject rate and the error rate of2.13%and1.37%, respectively, proved theeffectiveness and feasibility of the feature extraction algorithm.