Research On Hand Gesture Recognition In Image Sequence

The using of hand gestures has become an important part of human computer interaction (HCI) in recent years. The ability of computer to visually recognize hand gesture is essential for future human computer interaction. However, vision-based recognition of hand gestures is an extremely challenging inter-disciplinary project due to following three reasons:Firstly, hand gestures are rich in diversities, multi-meanings, and space-time varieties; Secondly, human hands are complex non-rigid objects; thirdly, computer vision itself is an ill-posed problem.The maincontent of thisthesis are as follows:Regarding the handgesture detection, this thesis puts forward the new handgesture detection method which utilize YCbCr-HSV Double Color-Spatial Model and Background Model algorithm to solve the problem that Single Kin-Color Model is obstructed by near kin color. This segmentation method was realized by the fusion of muti feature.In the part of hand gesture tracking, this thesis nalyses the Camshift algorithm. Through the analysis of the shortage, Camshift algorithm was improved according to the characteristics of the gestures tracking.this thesis puts forward the method that Camshift algorithm Combined with frame difference method and the Kalman filter. First of all, the target information is added to the calculation of color probability distribution by frame difference method. The automatic initialization tracking is realized and the colors that similar to the target in the background are rejected. When gestures fast motion and keep out leads to tracking accuracy low or track failure, utilize Kalman filter forecast the location of the object in the process of searching window. Call Camshift algorithm of target tracking iteration to locate and restore tracking. So as to realize the change of the automatic tracking by semi-automatic trackingRegarding the handgesture recognition, Described in detail based on Fourier descriptors and moment invariants of the gesture characteristics extract, this paper puts forward the static gesture recognition method based on Hu invariant moments and Fourier descriptor feature combination.Considering the accuracy of gesture modeling is inconsistent with its efficiency,a dynamic hand gesture modeling and real-time extraction method is proposed. It uses starting hand shape, end hand shape and middle trajectory as dynamic gestures model, using boundary moment to extract dynamic hand gesture features. Experiments show it can improve the speed of the dynamic gesture recognition.