Abstract: In order to solve the problem that it is difficult for the current image target localization algorithm to automatically locate the target to be deformed and the positioning time is long and so on, an automatic target localization algorithm based on sparse active contour model is proposed. The image of tortoise rubra is divided into blocks, and the shape of each block is estimated. Under the constrained target shape estimation, the common contouring algorithm is used to learn the target contour model, and the area with the highest matching value is scanned in the to-be-positioned map. Positioning results, the use of distance-based constraints Huff transform, accurate positioning results. The experimental results show that the proposed algorithm can effectively overcome the problem of large changes in the target size of the image, improve the automatic positioning accuracy of the target and reduce the time-consuming positioning.