Research On Key Technologies Of Two-Dimensional Cartoon Motion Capturing Based On Shape Matching

Two-dimensional cartoon animation is widely used in many fields, such as entertainment, advertisement, education, etc., because of its distinctive charm. How to reuse the existing cartoon animation effectively is an attractive research issue. Cartoon motion capturing is the base of cartoon animation reuse. This article does researches on some key techniques of cartoon motion capturing. The main contents are as follows:This article extracts binary area of cartoon character using background subtraction method based on single Gauss background model and YUV double threshold. A background subtraction based method of extracting pose sequence of cartoon character is designed: after getting the binary image of the character area, it is used to do and operation with the gray scale image of the origin to get pose of the cartoon character.This article proposes a method of simplifying binary image of contours of cartoon character. Morphology open operator, Fourier descriptors and polygonal curve approximation techniques are used in this article to remove unimportant detail information. As experiments show, matching accuracy of two contours of characters of adjacent frames is improved. Thus, the accuracy of track is improved.Key points of the cartoon character of the first frame are marked. Shape context is used to do shape matching between contours of characters of two adjacent frames. Thin plate spline is used to compute deformation, which is applied to key points to get their position after deformation. At last, the results of the capture are redrawn in OpenGL as a stick man.