Nowadays, many kinds of factors related with illumination in the gathering process of iris recognition system bring light spots to the iris images, and the light spots lead to the loss of the iris texture. To a great extent, it affects the precision of the iris recognition system. In order to improve the accuracy and practicality of iris recognition system, an iris recognition method based on eliminating disturbance points in iris image is proposed. This method uses the image fusion technology to match two images of the identical iris which have light spots in different positions. As a result, the light spots are effectively eliminated after the image fusion. Then it uses an iris block-encoding method based on statistic of local information for feature extraction and encoding. After that it calculates the Hamming distance between different iris codes and obtains the recognition results according to Hamming distance. When the Hamming distance threshold equals to 0.255, the accuracy of iris recognition is 99.74%. The experimental results demonstrate that this method can effectively eliminate the disturbance points and has high accuracy, which can meet the demands of iris recognition.