Pose estimation is based on the critical point analysis

摘要

Topic Method and the system of pause presuming object ones are offered.Solutions Any object things the human body and the animal, are good even with the robot. The depth information regarding object ones 310 with camera 110 is received, in pause presumption module 120 the pause of object ones and conduct is decided from the picture, the response for the pause and the conduct which are perceived in interaction module 130 is output. Due to pause presumption module 120, it divides the picture which includes object ones, makes the taxonomic part and the not yet taxonomic part. Segmentation making use of k-means cluster method is good. It is possible to pursue between the pictures the taxonomic part, as a known object of the head and the body section etc. Paralleling to x axis and the Y-axis, it scans the not yet taxonomic part, minimum value and maximum value specifies. The critical point is led from maximum value and minimum value 320. Tying the various critical points, although the latent joint it specifies, is suitable to the actual object of object ones the joint which possesses sufficient probability is selected, skeletal structure is formed on the basis of the pause of object ones of the decision being completed. Selective figure Figure 3