关键词: 虹膜识别, 特征提取, 方向能量特征, 旋转平移不变性

Abstract: In order to solve the problem that the traditional iris recognition system’s performance isn’t good enough under non-ideal image conditions, a novel algorithm based on texture energy features is proposed. Firstly, a set of horizontal and vertical direction filter is designed to extract iris texture edge. The characteristic figures exhibiting the differences of direction energy are generated by comparing the energy intensity of iris texture edge in the two directions, horizontal and vertical. Secondly, the figures are divided into different parts, and the maximal point of every energy difference is selected as an effective point to obtain eigenvector. Besides, swelling the noise template and ruling out the influence of noise point and convolution noise point on the surrounding effective figures. Finally, the hamming distance matching vector is obtained to implement iris recognition. The method is tested by CASIA3.0 iris database provided by Chinese Academy of Sciences gaining a good recognition rate.

Key words: iris recognition, feature extraction, directional energy characteristics, rotational and translational invariance