The Meaning Of The Gabor Transform Parameters And Their Effects On Iris Recognition

Biometric technology uses its own characteristics as identity markers, rather than usingthe traditional markers, so it is more secure, more portable and not easy to lose or fraudulentuse of features. The invariance and non-invasive which are as important identitycharacteristics make people generates a lot of interest on identity verification based on irisrecognition. Because of the similarity between2D Gabor function and human visual cortexcell receptive-field, feature extraction algorithm researched based on2D Gabor filter has beenmore and more popular.This paper focus on the significance of2D Gabor filter parameters used for featureextraction and the influence of various parameters values on the recognition rate. Firstly, thispaper described how to use2D Gabor filter to extract the iris texture and the significance ofthe2D Gabor function parameters. Secondly, it researched the Gabor filter parameters rangeas well as the influence of various parameters values on the recognition rate.Finally,according to the experiment on a large number of samples in CASIA iris database andJLU-IRIS iris database, proved the influence of the various parameters values on theeffectiveness and recognition accuracy of the feature extraction, as well as to verify theeffectiveness of the algorithm.The main contents of this paper relates to the following aspects:1) The significance and nature of the function parameters of2D Gabor function. Firstly,this paper analyzed the origin of2D Gabor function and its advantage in iris recognition.Secondly, it researched the various parameters specific meaning of frequency and direction of2D Gabor filters, illustrated the specific meaning of each parameter. Finally, it elaborateddirection, frequency and local characteristics etc.2) Researched the various parameter values of the2D Gabor filters using for extractingiris texture information. This paper first analyzes the particularity and diversity of iris textureinformation. Then studied a reasonable range of parameter of2D Gabor used to extract iristexture information, including the parameter of frequency, octave, scale and direction. Finallyselected15categories of images from the CASIA iris database and9categories of images from the JLU-IRIS iris database, optimized the values of various parameters using geneticalgorithm to obtain a reasonable value set of parameters of the filter to extract the texturevalid information.3) Researched the influence of various parameters values on the recognition rate. Donefeature extraction and pattern recognition to70class images in CASIA iris database and9class images in JLU-IRIS iris database based on different values of2D Gabor filterparameters. The different parameters have a great impact on the consequent, the recognitionrate reached92.8%and97.3%respectively under the conditions of the best combination ofparameters. The experiments also proved the effectiveness of the2D Gabor filter to extractthe characteristics, the influence of various parameters on the recognition rate was analyzed,and in the direction and the other characteristics of the2D Gabor filters.In summary, this paper researched on the significance of2D Gabor filter parameters, andits nature as well as the values of the parameters used to extract the iris texture information.Proved the characteristics of2D Gabor filter, in addition to influence of parameters whichwere set different values on iris recognition rate by experiments, and improved recognitionaccuracy of the iris recognition system of the laboratory.