In view of adaptive ant colony optimization (AACO) algorithm in image edge extraction often appears problems of low efficiency, and easily falling into the local extremum. An adaptive ant colony image edge detection algorithm combined with biological Predator-Prey behavior is proposed. The algorithm is combined with AACO algorithm and Predator-Prey behavior. Ant colony is divided into Predator and Prey species, AACO algorithm is used to search the two species in the initial stage, after a certain number of iterations, the two species enter the exclusion stage. The method of automatic threshold is used to extract image edge. Experimental results show that compared with AACO algorithm and Canny algorithm, in terms of accuracy, image edge extrated by the proposed algorithm is better than the edge extracted by the first two algoritms. Meanwhile, it maintains fast convergence characteristic of AACO algorithm, and overcomes its shortcoming of easily falling into local extremum. So image edge can be effectively detected by it.