Dynamic auto focus algorithm based on improved Sobel operator

Aimed at improving performance of automatic focusing algorithm in dynamic environment, this paper describes recent studies of focusing evaluation function and search strategy. On the basis of analysis of characteristics of human visual system, new focusing evaluation function based on eight direction Sobel operator edge weighting is proposed. At the same time, adaptive variable step search strategy is used in order to overcome the disadvantage of slow speed of traditional climbing method. Simulation experiment results show eight direction Sobel edge detection operator has good edge detection effect, and focusing evaluation function has better anti-interference ability than traditional two direction Sobel operator focusing evaluation function, which is calculated by this operator combined with characteristics of human visual system giving the edges with different weight coefficients. Finally, an experimental platform based on liquid lens is set up, which verifies the performance of improved auto focus algorithm in dynamic environment. Experimental results show focusing accuracy with proposed algorithm can achieve 97.5% in dynamic environment.