Research On Human Skeleton Extraction Technology Based On Depth Image

Human motion analysis is one of the important research areas of human computerinteraction, human capture is an important part of human motion research, due to the skeletonextraction technology is the key step in the human body movement, the research of skeletonextraction method has also become a hot topic in the field. Human skeleton carries a smallamount of information which can display all kinds of human action effectively, and can alsoimprove the processing speed of the computer. In this paper, by using Kinect hardwaredevices for depth image, the human skeleton extraction technology is analyzed based on thedepth of the image studying.The paper instructs the background and significance of the topic, as well as the relativetechnology of Kinect and depth images. Firstly, the median filter is chosen to process thedepth images, which are obtained directly, then use thresholding on the depth image to extractthe foreground from the image. Continually, using the morphological transformation methodincluding dilation, erosion, open operation and close operation to removal the image borderand internal noise points, as well as does the comparing and experimental analyzing theforeground segmented of depth image result and the traditional frame difference result. In thestudy of skeleton extraction method, respectively researched and implemented methods indistance transform including Euclidean distance, City block distance and Chessboard distancemethods, thinning algorithm including Zhang parallel fast thinning algorithm and Hilditchthinning algorithm, experimental results show that Zhang thinning algorithm of the humanskeleton extraction is better. Simultaneously, this study use the feature of depth image withdepth information which can solve the human self-occlusion issues. According to NASAanthropometric proportion position of the human skeleton points in the skeleton line, the headcenter of gravity is positioned as the head skeleton joint position, and uses the largest trianglemethod to get the elbow joint point. Finally, the experimental results illustrate theeffectiveness of the skeleton extraction method.