None of the traditional methods for extracting contour has the good self-adaptive ability. To solve this problem, a method based on ant colony optimization algorithm was presented for extracting seed’s contour in digital image. The digital image was taken as a two-dimensional grid. The seed’s contour was the food for ant. The gradient and direction in the preceding step were the heuristic information which guided the ants to search the real contour together with pheromone intensity. Reasonable heuristic information and pheromone update mechanism was defined to increase the speed of convergence and raise the search ability. The relationship between the parameters in the algorithm was also studied. At last, the acquired images under two different kinds of illuminations were used to test the algorithm. The experiment result showed that the algorithm not only has nice robustness and parallelism, but also has preferable self-adaptive ability, and extract contour more accurately than the traditional methods. 