摘要 该文提出了一种工作于MPEG压缩域的快速视频目标分割算法。该算法以从MPEG1/2码流中部分解码提取的特征为输入,提取P帧中的运动目标。针对一般的压缩域算法目标边界精度不高的特点,算法采用I帧和P帧中每个块的直流DCT系数和3个交流DCT系数,以及运动补偿信息,重建出P帧的原图像1/16大小的子图像,采用快速平均移聚类得到具有较高边界精度的亮度一致的区域;针对运动向量的噪声容易造成错误检测的缺点,算法结合聚类分析结果和运动块的分布,采用基于马尔可夫随机场的统计标号方法对目标和背景区域进行分类,得到每个P帧的目标掩模。该算法可以得到4×4子块的边界精度,对于CIF格式的码流,在Pentium IV 2GHz平台上可以达到每秒40帧的处理速度。
Abstract:A fast video object segmentation method working in MPEG compressed domain is presented in this paper. Moving object masks in P frames are extracted by exploiting features obtained by partial decoding. To increase object boundary precision, for each P frame, a 1/16 sub image is constructed using DC and three AC coefficients, and motion compensation information, then a fast mean shift clustering algorithm is used to divide the image into regions with coherence luminance and obtain high precision region boundaries. For reducing the influence of motion vector noise, a MRF-based statistical labeling method is exploited to classify regions into two classes: moving object and background. The proposed algorithm can get a boundary precision of 4×4 sub-block with a high processing speed. For CIF video streams, the algorithm can run at a speed of 40 frames per second in a Pentium IV 2GHz platform.