Research On Palmprint Recognition Algorithms Based On SIFT

With the development of social economy and science technology,people rely on identity information gradually for their work,life,and travel.Palmprint recognition has emerged as a new biometric technology.Palmprint contains rich individual identity information.Its features are stable and unique,and have strong anti-noise capabilities.In recent years,palmprint recognition become an important research object in the field of pattern recognition and human-computer interaction.Through reading a lot of documents about the palmprint recognition,analyzing the status of palmprint recognition algorithms at domestic and abroad,and existing tradition algorthms are sunk in.In this paper,two types of palmprint recognition algorithms based on SIFT of feature points are designed and studied,which combine HOG and LBP.The main works of this article are as follows:?In view of the fact that the SIFT has fewer characteristics of actual discrimination ability,and the dimension is higher,the feature point description sub-generation is more complicated.The HOG palmprint recognition algorithm based on the largest rectangular area of SIFT feature points is designed.In the palmprint region of interest,SIFT is used to extract feature points,and the largest rectangular region is selected based on the maximum and minimum coordinates of the feature points to extract its HOG features,and the nearest neighbor method is used for classification and recognition.Based on the PolyU palmprint library,simulation experiments and analysis are performed.Experiments show that the HOG algorithm based on the maximum rectangular area selection of SIFT feature point effectively reduces the dimension of the SIFT feature.And the recognition performance of this algorithm has also been improved.?Designing LBP palmprint recognition method based on SIFT feature point detection can effectively reduce the dimension of the feature vector,and LBP is more targeted when selecting global features.SIFT is used to extract feature points in the region of interest.The surrounding neighborhood selection method is designed basedon the feature points.In order to maintain the rotation invariance of the feature points,the image is rotated in the reference direction.The rotation-invariant LBP feature is calculated with the feature point as the center.In the feature description stage,weighted processing is performed on the pixels within the neighborhood of the feature points,and all the weighted LBP features are combined and normalized.The images are classified using the nearest neighbor classification method to obtain the recognition result.The experimental results confirm that the weighted processing can improve the recognition performance of the algorithm,and the LBP feature recognition based on the SIFT feature points surrounding neighborhood selection.This algorithm is superior to the recognition effect of a single algorithm and confirms the effectiveness of the algorithm.