The invention further discloses a tuning peg transformation-based feature region identification method for pigs. According to the method, firstly, the original image of a pig which is collected in theoverlooking condition is pre-processed, and the target contour image of each required single pig is obtained by applying the morphology-based prospect detection and target extraction algorithm. Secondly, based on the harris corner point detection algorithm, each salient key point is obtained. On the above basis, the delaunay triangulation is carried out. By means of the method of tuning peg transformation, a pig target is divided into a plurality of segmented regions, wherein the segmented regions around the contour of the pig are called salient regions. Meanwhile, the relationship between the segmented regions is represented by a tree structure. Finally, a multi-classification method based on a support vector machine is combined to identify the salient regions. Therefore, all important feature regions (head, ear, nose, leg, tail) of a target object are obtained, so that a new idea is provided for the identity and behavior recognition of pigs.