视频对象的分割是基于内容的视频处理的重要组成部分。提出并实现了一种基于水平集的运动视频对象分割算法。算法通过视频帧间的亮度差值提取初始轮廓曲线,将该曲线作为水平集算法的初始零水平集,采用窄带水平集方法演化曲线。得到最终的分割结果。实验表明该算法简单高效,具有很好的分割效果。
Video object segmentation plays an important role in the video process. A motion object segmentation algorithm based on level set method is proposed and implemented. The algorithm gains the primal contour curve by luminance difference between the video frames, and sets the curve as the primal zero level set. Then the narrow band level set algorithm is used to evolve the curve until achieving the segmentation result. The algorithm is speediness and efficiency, and can achieve a good segmentation effect.