In order to reduce noise influence and improve false contour phenomenon caused by automobile body reflection, a new contour extraction algorithm combining visual neurons dual antagonism mechanism and adaptive dual threshold is proposed. Original images are processed by dual antagonism mechanism algorithm to get contours of brightness characteristics according to visual neurons dual antagonism principle. High and low threshold values are selected automatically according to brightness distribution of the contours by using extended Otsu algorithm. Double threshold processing is carried out to the contour preliminary extracted, and the final automobile contour is obtained. The objective performance index is used to evaluate the experimental results. Results show that the proposed algorithm improves the integrity of automobile contour. Compared with the classic contour extraction algorithm, it has low computing complexity and good adaptability of image background.