Study On Palmprint Recognitation Based On2D And3D Palmprint Image Orientation Features Fusion

Traditional palmprint recognition are mostly based on2D palmprint image. Though by this way can achieve a good result, there are also many problems, such as the recognition rate is easily affected by the illumination, palm oil and other factors. Based on the latest achievements of research at home and abroad, this paper research on the2D and3D palmprint information fusion methods and fusion images recognition, the main work is as follows.Firstly, the3D palmprint obtained based on structured light acquisition system containing rich spatial structure and direction information. Take full advantage of the3D palmprint mean curvature image information. Using the line features and direction features, recognition based on the weight matching of decision level and feature level fusion. The experimental results show that, the methods have higher recognition rate.Secondly, according to the characteristics of2D palmprint recognition is influenced by the external environment, and based on the two-dimensional Gabor filter in the superior characteristics of extracting texture direction information. We put join the3D palmprint mean curvature information, using the method of matching weight fusion at the decision level. The experimental results show that, this method can effectively improve the recognition rate and enhanced robustness.Finally, based on the3D palmprint contains the information of the spatial structure which the2D palmprint does not have, the proposed methods, respectively, using a two-dimensional discrete wavelet transform and multi-scale and multi-direction Contourlet Transform for2D palmprint image and3D palmprint MCI image decomposition and fusion. Then on the basis of the fusion image using Gabor filter to feature extraction, use of competition code coding rules matching and identification the extraction of the characteristics. Experimental results show that, compared with simple2D palmprint recognition and3D palmprint recognition based on Gabor filter feature extraction, the proposed methods have higher recognition rate. The methods also have higher security and robustness. From actual needs, the proposed methods have higher efficiency, so the methods can meet the practical needs.