Research On Iris Feature Extraction And Recognition Based On The Adaptive Gabor Filtering

Human society is now in a stage of rapid development of information technology to permeate every area of people’s lives, the attendant a lot of information security problems. social urgent need for an identification method of safety and effective. Biometrics is a means to rely on human beings biometric feature, to define user access and operating authority, to solve the current society facing the issues of information security. Compared with other biometric,iris feature with its unique, security, non-invasive and so on,more and more be paid attention from the field of security scholars. Currently iris recognition products have been used in access control systems, the financial sector, airport security and other places that require a high degree of security, but due to the diversity and complexity of its practical application environment, there are still many places to be perfected.In this paper, the target is to improve the practical application performance of iris recognition. mainly study the latter part of the process of iris recognition, that the part of feature extraction and feature recognition. To achieve the iris recognition system can accurately identify in a complex condition, feature extraction using Gabor parameter optimization algorithm based on chaos particle swarm,and achieve adaptive Gabor filter to extract phase and direction of the two iris texture feature. In feature recognition, the matching results of phase feature and direction features of iris as input data of SVR, the use of SVR model training iris template feature to identify and test samples.The main work includes the following aspects:(1) In the iris image acquisition process, the interference by eyelids and eyelashes is inevitable. In this paper, we do preproccess for collected the original iris image. We intercept the prominent part of iris texture features as region of interest, as to improve quality of input image in iris recognition system.(2) In regard to feature extraction methods, this paper proved that the effects of different Gabor filter parameters for iris recognition results by experiment. Therefore, in order to improve the accuracy of iris recognition, this paper proposed using CPSO optimization algorithm to optimize parameters of Gabor filter,and prevent premature convergence to local the max value adding a chaotic particles in each iteration. The distinction degree of intra-class and inter-class is regarded as the fitness function of Chaos particle swarm algorithm.And constantly updated Gabor parameters and maximum fitness value, until finding the optimal value or reaching the number of iterations.(3) In this paper, SVR fuses the matching results of the phase feature and orientation feature of Gabor.Using "0" and "1" respectively identify two subsets of intra-class and inter-class. Iris classification model obtained by SVR training, and use this model to identify the test sample. Experiments show that extracting two iris features can reduce the instability of recognition caused by a single feature,and further improve the accuracy of iris recognition.In this paper, experimental data use the CASIA image library CASIA-V4-Interval, CASIA-V4-Lamp and laboratory independently research and development iris capture device to capture an iris image of JLU-V3.0 libraries, and using distinction degree of intra-class and inter-class to evaluate performance of iris recognition method.In summary, this paper proposed chaos particle swarm optimization to optimize Gabor parameter. Providing the iris recognition method based on adaptive Gabor filter, and through SVR model fused the matching results of two feature of iris. Finally, by contrasting experiment verify the effectiveness of the algorithm.