一种视频对象生成方法的研究与实现

Study and Implementation of a Video Object Generation Algorithm

提出了一种视频对象的自动生成方法。首先对当前帧进行全局运动估计和补偿 ,并和上一帧进行差分运算 ,然后采用模糊分类器自动检测出运动区域 ,再结合边界特征形成视频对象的二进制模型并生成视频对象。实验结果表明 ,该方法能有效地生成视频对象 。

This paper presents an automatic generation algorithm for video objects. First, the current frame is global motion compensated after the global motion estimation, and the difference operation is made between the current frame and the previous frame. Then, motion region can be detected by using a fuzzy classifier and the binary models of video objects can be determined by combining motion region and edge information in the current frame. Finally, the video object can be generated according to its binary model. Experimental results show that the proposed algorithm is effective with strong robustness.