摘要 该文提出了一种对视频序列中的运动目标进行自动分割的算法。该算法分析图像在L U V空间中的局部变化,同时使用运动信息来把目标从背景中分离出来。首先根据图像的局部变化,使用基于图论的方法把图像分割成不同的区域。然后,通过度量合成的全局运动与估计的局部运动之间的偏差来检测出运动的区域,运动的区域通过基于区域的仿射运动模型来跟踪到下一帧。为了提高提取的目标的时空连续性,使用Hausdorff跟踪器对目标的二值模型进行跟踪。对一些典型的MPEG-4测试序列所进行的评估显示了该算法的优良性能。
Abstract：A new automatic video sequence segmentation algorithm that extracts moving objects is presented in this paper. The algorithm exploits the local variation in the L~*u~*v~* space, and combines it with motion information to separate foreground objects from the background. A new image segmentation algorithm based on graphic-theoretic approach is first employed to generate various regions according to local variation. Next, moving regions are identified by a new filter criterion, which measures the deviation of the estimated local motion from the synthesized global motion. In order to increase the temporal and spatial consistency of extracted objects, moving regions are tracked by a region-based affine motion model. A two-dimensional binary model is derived for the objects and tracked throughout the sequence by a Hausdorff object tracker. The proposed algorithm is evaluated for several typical MPEG-4 test sequences. Experimental results demonstrate the performance of the proposed algorithm.
2002, Vol. 24 
