Abstract: In order to resolve the problem of recognition rate decrease in condition of unideal iris image and local noise，an iris feature extracting algorithm based on local zero-crossing detection is presented in this paper. This method extracts iris texture feature by using local relativity of zero-crossing operator and texture signal, then encodes iris texture to form iris feature template depend on coefficient sign, finally classifies different patterns using similarity. Simulation results show that stable feature of unideal iris image can be extracted and high recognition rate can be achieved using the algorithm.