Research Methodology Of Fuzzy Palmprint Feature Extraction

Because the palmprint have rich information and easy to access, palmprint recognition research becomes one of the hot areas in biometrics recognition recently.The palmprint images which are taken within depth of field, have most of the details, and are very suitable for large groups as identity recognition. In order to ensure the palmprint images are acquired within the depth of field, contact mode is used usually. Considering the possibility of disease transmission, some people reject to the contact mode. Therefore, this paper research on the contactless method to acquire palmprint images, and the testers are asked to put hands on the set position for shooting. Since each person's feeling about distance is different, some people may put hands in the position out of depth of field, and this will result in blurred images taken of the palm, which part of the texture detail is lost. When the extent of deviation from the focal plane increase, more texture information of palmprint image are lost, and resolution of texture information becomes lower. As the palmprint has rich texture information, it can have higher resolution even if there is a certain image blur. If curves of corresponding relationship between image definition and system identification rate are created, the clarity scope of palmprint image can be determined according to the system recognition rate. In fact, the allowable range between the hand and the camera can be further determined. In other words, on the occasion of the wrong recognition rate requirement is not very high, the distance range between the hand and the camera can be greater.This paper research on contactless palmprint image acquisition, gradient operator based on the definition evaluated function is introduced, the relationship between palmprint definition evaluated function and the shooting distance is discussed, and corresponding curve model is established. Then extract the palmprint features for evaluating, select the stability characteristics for feature extraction and recognition, and establish relationship curve model between the image definition and error recognition rate. According to the allowed error recognition rate by system and the relation model between the image definition and error recognition rate set image clarity threshold, and realize improved non-contact on-line palmprint recognition simulation system based on image definition evaluation module. The experimental results show that the improved simulation system ensures the effectiveness of the image acquisition, and reduces the error recognition rate and has practical value.