Finger Vein Recognition Method Based on Center-Symmetric Local Binary Pattern

The classical local binary pattern (LBP) comes from a single circular region or a square region, so it is impossible to express the correlation between multi-neighborhood pixels under the same central pixel condition. When the LBP histogram is used to describe the partitioned area, the dimension of the histogram is high, and the space and time consumption is too large. In order to overcome the shortcomings, the center-symmetric local binary pattern (CSLBP) is adopted in this paper. Firstly, LBP and CSLBP operators are studied and analyzed. The binary feature coding of the finger vein by using LBP and CSLBP are extracted. Secondly, the experimental preprocessing of the finger vein image is done. LBP and CSLBP operators are used to extract the finger vein feature information to obtain two texture feature images. Finally, the features of finger vein images are matched by the histogram intersection kernel. The experimental results show that the background modeling using CSLBP can effectively solve the dimension disaster and noise sensitivity problems caused by the LBP operator. The detection effect and the calculation speed are improved.