Manual Shape Parameter Detection, Static Digital Image

In the domain of computer vision, there is a very wide application for the research of human motion analysis. And there are more and more researchers and institutions get focus on these area. Human hand motion analysis, as a branch of human motion analysis , can be used widely in helping deaf people communicated with computer and high level HCI(human computer interaction ). However, because human hand motion is highly articulated and self-occlusion , most hand motion analysis are highly depended on CyberGlove. Despite of the convenience and accurate CyerGlove have, the high price prevent its become a common used hardware equipment in lab.A number of papers by using CyberGlove have found that human hand motion is correlated between and within the fingers. Moreover, one can predicted hand pose in self-occlusion situation through pri-knowledge of hand motion constraints.The research is funded by Shanghai natural science fund. We describe an approach for human hand motion detection using an articulated model with hand kinematics constraints from monocular view.First we introduce an initialize gesture, favorable stretched-out hand pose image. And this gesture has no self-occlusion problem from which we get the width of each finger ,the joint position within fingers. Then we work with gesture image that need to be estimated , from which we can get part information un-occlusion . At last, combined the initialize gesture information and hand motion constraints we can get the whole information.Because the hand motion constraints was given by other papers. So we focus on how to get the full and accurate hand information from images. To achieve this we use a batch of algorithm, such as image binary , skeletonize, contour detection, ellipse fitting.