A new method of breakpoint connection for human skeleton image

There are many discontinuous skeleton points in human skeleton images, which make human skeleton behavior analysis be a difficult problem. Our paper presents a new breakpoint algorithm based on layer and partition of the neighborhood. We scan skeleton images line by line. The number of other skeleton points is calculated for each skeleton point in its 8-neighborhood. Then we judge whether this skeleton point is a breakpoint or not according to the number of the above-obtained and the distribution of other skeleton points in its 8-neighborhood. If it is, we will find available connection skeleton points which can connect the breakpoint. Finally, we find out the points that need to be updated to complete the breakpoint connection progress in accordance with linear equations established by the breakpoint and available connection points. Through of theory analysis and experiment verification, our method has good effect on connecting breakpoints in skeleton images and the shapes of skeleton images are undeformed. In addition to the human skeleton images, this method can also be used for other objects skeleton images on the breakpoints connection.