基于SLIC和马尔科夫随机场的H.264视频对象分割  

H.264 Video object Segmentation Based on SLIC and Markov Random Field

作　　者：王美杰 张起贵 李付江 WANG Mei-jie;ZHANG Qi-gui;LI Fu-Jiang(School of Information and Computer,Taiyuan University of Technology,Jinzhong Shanxi 030600,China)

机构地区：[1]太原理工大学信息与计算机学院,山西晋中030600

出　　处：《计算机仿真》2021年第7期170-175,共6页Computer Simulation

基　　金：山西省基础研究项目自然科学基金(2013011017-3);国家自然科学基金资助项目(61471255)。

摘　　要：为了提高H.264压缩域视频对象分割时的鲁棒性和准确性,提出了一种基于简单线性迭代聚类(SLIC)和图割优化的马尔科夫随机场(MRF)运动对象分割算法。算法直接利用从摄像机产生的H.264压缩码流中提取的运动矢量。首先对运动矢量场进行预处理,然后构建基于改进的SLIC分割的马尔科夫模型能量函数,最后利用图割法求解能量函数进而分割出运动对象。在公开的数据集上进行实验表明,与近年来经典压缩域视频对象分割算法相比,上述算法在复杂背景下可以有效提高分割的准确率和F度量,运算速度平均提高约1.85倍。与先进的像素域分割方法相比,运算速度提高了5倍,算法适用于实时性要求较高的视频监控场合,可有效减少数据存储和处理的内存需求。A Markov random field segmentation algorithm based on simple linear iterative clustering(SLIC) and graph cut optimization, was proposed to improve the robustness and accuracy of video object segmentation in the H.264 compressed domain. The algorithm mainly used the motion vector extracted from the H.264 compressed stream. Firstly, the motion vector field was preprocessed and then the Markov model energy function based on SLIC segmentation was constructed. Finally, the graph-cut method was used to solve the energy function and segment the moving object. Experiments on open datasets show that the algorithm can effectively improve the accuracy and F measurement of moving objects in the complex background, compared with the classical video object segmentation algorithm in the compressed domain in recent years, the operation speed can be increased by 1.85 times on average. Compared with the advanced pixel domain segmentation method, the operation speed is improved by 5 times. The algorithm is suitable for video surveillance with high real-time requirements, and it can effectively reduce the memory requirements for data storage and processing.

关 键 词：压缩域 简单线性迭代聚类 视频对象分割 马尔科夫随机场 图割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]