The Research On Iris Location And Recognition Algorithm

Because the iris recognition technology is unicity, stability, collected, noninvasive and high-security, it has been developed rapidly. However, at the present time, the iris recognition technology is not used widely, the chief cause is that the precision and speed of iris positioning and the effectiveness of iris feature extraction have influence on the practicability of iris recognition technology.In this paper, we focus on the research of the positioning algorithm and feature extraction algorithm in the preprocessing. The paper is outlined as follow.In iris positioning:In order to improve the precision and speed of iris positioning, in this paper, we adopt a wavelet-based positioning algorithm. At first we decompose the image in use of wavelet transformation, then extract the iris inner and outer boundary of the low-frequency part of the transformed image.For the purpose of solving the positioning problem of iris outer boundary blurring contour, in this paper, on the basis of a priori knowledge of iris image database, we introduce a positioning algorithm, this algorithm improve the blindness of searching and the positioning speed by reducing the searching area step by step. In use of the method of median filtering and canny operator parameter set-up, we get the clear contour of outer boundary. At last, we compare the method with the two traditional positioning method respectively. The algorithm proposed in this paper can short the positioning time in the premise of ensuring the accuracy. The second method in our paper has high accuracy of outer boundary positioning.In feature extraction:For the existing algorithm only takes into account radial information but ignores the angle information when extracts the texture feature. In this paper, we adopt the method based on 2-D Log Gabor Filter to extract the texture feature of iris, and it takes both radial and angle information into account and extracts the iris texture feature more effective. In addition, we analyze deeply the effect on accuracy of feature extraction when we adopt different parameters of filter.We match the iris feature vector by hamming distance. Finally we compare the algorithm proposed in this paper with recognition method based on Daugman algorithm. The experiment results show that the algorithm we proposed can increase the recognition speed in the premise of ensuring the higher accuracy.