A new method for iris recognition is presented, which can help users to get correct position and accomplish non-contact capture by self-aiming. The experiments on extracting feature from iris texture are carried out, and the feasibility and efficiency of this method are verified. The special illumination is designed for the purpose of assisting users to focus their eyes on the proper position. A piece of pellicle mirror is placed in front of the camera lens to help users to observe the eyes' image and ensure that the eyes are in the camera visual field. Thus image data that meet the requirement of iris recognition can be obtained. The two-dimensional Gabor wavelet theory is adopted for extracting feature from iris texture. Feature matching experiments are made by counting the amount of the equal flag bits of two iris images.