作　　者： ; ; ; ;

机构地区： 浙江万里学院电子信息学院宁波市数字信号处理DSP重点实验室

出　　处： 《电路与系统学报》 2008年第2期122-126,共5页

摘　　要： 面向语义视频检索,提出一种压缩域的目标分割新算法。它直接基于压缩码流中运动矢量和DCT系数,经过运动检测、矢量分水岭分割、目标融合与修正、后处理与跟踪等步骤提取空时视频目标。整个过程主要基于压缩域进行,无需视频码流的完全解码。对不同测试序列的实验测试结果显示算法能基于压缩域提取较为精确的空时视频目标,并具有较好的鲁棒性。 A novel scheme for object detection and segmentation in compressed domain is proposed aiming at semantic video retrieval applications. The spatial-temporal object can be finally extracted after steps of motion detection, vector-based watershed segmentation, object fusing and correction, and finally post-processing and tracking. The whole process is mainly implemented in compressed domain without the need of fully decoding of compressed stream. Simulations are performed on several standard video sequences and the results reveal that the proposed algorithm can extract accurate spatial-temporal object directly from compressed stream effectively and robustly.

关 键 词： 视频检索 压缩域 空时视频目标分割 矢量分水岭算法

领　　域： [自动化与计算机技术] [自动化与计算机技术]