【摘要】：Human motion analysis, one of the front porch orientation drawing much attentions in recent years, is an very active branch of image processing and computer vision territory , it is applied widely in many aspects of national economy and military fields. The research object in this thesis is video-based human motion, namely, tracking capturing; and analysing human motion in video. Human motion is non-rigid object motion and nonlinear. It has potential research value and application prospect.
Though using video to analysis human motion has been a challenging problem confronted by many researchers, some solutions to this problem are presented partially. The broad-jump players in the list eye video frequency are set to be the research object, the present video-based human motion analysis technique is introduced and a human motion analysis system is completed.
The following is the main task of this thesis.
(1) Introduce video array analysis basis, hat is, the implication of image, the visual principle of moving object, the enhancement rehabilitation smooth and filter of image, and the basis of maths morphology.
(2) Introduce the usual tracking methods and motion target test of video-based array, the instantaneous dividing methods can easily and accurately test the motion target, so this system applies instantaneous dividing method to test the divide human motion target. The application of Kanlman Filter to track the target makes clear the central motion of human body and the motion orbit of partial joints, some kinematics and kinetics parameter necessary for human motion analysis.
(3) Apply VC++ developing platform to complete video-based human motion analysis system, the experiment-testing and results-analysing proves that the system can offer a better analysis on human mation.