Three-dimensional Human Motion Color Video Segmentation Method

Human motion analysis is receiving increasing attention from computer vision researchers. This interest is motivated by applications over a wide spectrum of topics. For example, segmenting the parts of the human body in an image, tracking the movement of joints over an image sequence, and recovering the underlying 3D body structure is particularly useful for analysis of athletic performance as well as medical diagnostics. The thesis is divided into 5 sections:(1) Incorporate three methods using the VC++ language and get a 3D human motion analysis system. The three methods are: correlation tracking, camera calibration and 3D motion measurement.(2) Two experiments for this system are done: One is about system accuracy testing, Another is label-based human motion analysis.(3) In active contour based tracking, segmentation and tracking of video moving object is used for guiding the extraction of video object plane from the video sequence. This dissertation describes spatio-temporal information to automatically segment and track video moving objects. Firstly, a fourth-orders statistic hypothesis testing in inter-frame difference is used to automatically separate the moving areas from the background in a general video sequence. Secondly, Change areas are extracted as markers of the corresponding moving object. Using the marker, region-growing algorithm can locate the precise boundary of moving object.(4) In feature-based tracking, a histogram-based tracking algorithm which finds a probable location of the interested object is introduced. Then improves QTSU, introduces a method which segments an image by the center of shape of histogram and compares the two methods. Finally, tracking the feature of the object by the center of gravity and giving the tracking curves.(5) All the method has been programmed and the result been given and analyzed.