Research On Palmprint Recognition Algorithm Based On Non-Gaussian Gabor Filter

In recent years,with the rapid development of information technology and network technology,information security has shown an unprecedented importance.Biometrics technology is more and more widely used for its unique stability,uniqueness and convenience.Palmprint recognition technology as a new biometric identification technology,is one of the hot spots in recent years.Palmprint is used to describe the internal surface of all the lines on the palm of the collectively,it includes the main line,wrinkles and masturbation and so on.Compared with other biometric methods,the palmprint feature information is more abundant,the image acquisition is simple,easy to implement and accepted by the user,the recognition speed is quick,and it is easy to integrate with other biometric methods.Therefore,palmprint recognition is a very promising method of identity authentication.Palmprint recognition After more than 10 years of research,in theory and practice have a certain basis.In the palmprint image preprocessing tends to mature.As the palm surface of the surface lines of different shades,the direction of irregular changes,and lead to palmprint image changes and deformation of the more factors,seriously affecting the accuracy of identification.In view of the above problems,this paper takes the palmprint image as the research object and studies the existing palmprint recognition algorithm.This paper presents a new method of palmprint recognition with high accuracy and stable performance from palmprint feature texture analysis,feature fusion and spatial dimension reduction.The main research is as follows:1.The existing palmprint recognition preprocessing algorithm is studied and analyzed,and the region of interest(ROI)of palmprint image is extracted,and the selected algorithm is compared.2.A palmprint recognition algorithm based on non-Gaussian Gabor filter is proposed.We have constructed a function of non-Gaussian Gabor filter.We do the non-Gaussian Gabor filtering process on the preprocessed palmprint image.In this paper,we will recalculate the phase matrix of the filtered image and block the phase matrix.And then the histogram feature vector of each sub-block is connected,the histogram feature vector of the palmprint image can be obtained,and finally the vector is sent into the classifier,and then the vector of the LBP histogram is extracted.Classification and identification;3.In the Polyu-Pamprint-Database,and other palm prints were tested,and compared with other mainstream algorithms,the proposed method to achieve the desired effect,that the proposed method can effectively carry out palmprint recognition.