Abstract:At present, the study on people recognition is still a quite challenging problem, and it is a new subject to combine with multi-pose for person recognition. Therefore, it is a significant step for people recognition to capture multi-pose samples accurately. Poselets can detect all the people and the corresponding position in the image, but it cannot locate the people on a particular location. Therefore, this paper presents a method of pose extraction which based on poselets for the specific location:First, we should set filtering model according to the head calibration form of the human in specific location, and then using the filtering model, we can screen the figure box detected by the poselets algorithm and sort the results of the screening. Then, we can find the target person of the specific location by combining the sorting score with the maximum weight bipartite graph matching algorithm. Finally, it is easy to extract the corresponding pose. Experimental results show that the algorithm mentioned in this paper can detect the person of the specific location effectively and extract the corresponding pose.