基于形状匹配的二维卡通运动捕捉关键技术研究

发布时间：2018-03-10 12:50

本文选题：卡通运动捕捉　切入点：双阈值　出处：《天津大学》2012年硕士论文　论文类型：学位论文

【摘要】：二维卡通动画因为具有其独特的魅力而被广泛应用于娱乐、广告、教育等领域。如何有效地对已有的卡通动画进行重用,是当前倍受关注的研究课题。卡通运动捕捉是卡通动画重用的基础。本文就卡通运动捕捉中的若干关键技术进行研究,主要内容如下: 综合使用基于单高斯背景模型和基于YUV双阈值的背景减除方法提取卡通角色的二值区域。设计了一种基于背景减除的卡通角色姿态序列提取方法:得到角色区域的二值图像后,将其与原图像的灰度图像做与运算,得到卡通角色的姿态图像。提出了一种对轮廓二值图像进行形状简化的方法。使用形态学开运算、傅里叶描绘子和多边形曲线逼近等技术对卡通角色的轮廓进行简化,去除非重要的细节信息。实验证明,本文提出的方法可以提高两帧中角色轮廓的匹配精度,从而提高关键点的跟踪精度。首帧标定卡通角色的关键点,使用形状上下文对连续两帧中卡通角色的轮廓进行形状匹配,应用薄板样条函数计算变形,并将变形应用于关键点求解,获得它们变形之后的位置,最后在OpenGL中以火柴棒模型重绘捕捉结果。
[Abstract]:Two-dimensional cartoon animation has been widely used in entertainment, advertising, education and other fields because of its unique charm. Cartoon motion capture is the basis of cartoon animation reuse. In this paper, some key technologies in cartoon motion capture are studied, the main contents are as follows:. The binary region of cartoon character is extracted by using background subtraction method based on single Gao Si background model and YUV double threshold. A method of extracting cartoon character pose sequence based on background subtraction is designed: the character region is obtained. After the binary image of the. The gesture image of cartoon character is obtained by doing and calculating the grayscale image of the original image. In this paper, a method for shape simplification of binary contour images is proposed, which simplifies the contour of cartoon characters by using morphological operations, Fourier descriptors and polygonal curve approximation techniques. Experimental results show that the proposed method can improve the matching accuracy of role contours in two frames and thus improve the tracking accuracy of key points. In the first frame, the key points of cartoon character are calibrated, the shape context is used to match the contour of cartoon character in two successive frames, the deformation is calculated by using thin plate spline function, and the deformation is applied to solve the key point. The deformed position is obtained and the results are repainted in OpenGL with a matchstick model.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP391.41

【参考文献】