一种新的复杂场景中动目标图像分割的快速算法  

A NEW FAST IMAGE SEGMENTAION ALGORITHM FOR MOVING OBJECT IMAGED IN COMPLEX BACKGROUND

作　　者：李晓春[1] 孙培懋[1] 金国藩[1] 刘上乾[1] 

机构地区：[1]清华大学精仪系,西安电子科技大学

出　　处：《红外与激光工程》1996年第5期30-35,共6页Infrared and Laser Engineering

摘　　要：一种利用序列图像从复杂背景中分割动目标图像的新算法：首先利用马尔科夫随机场（ＭＲＦ）模型和连续线过程（ｃｏｎｔｉｎｕｏｕｓｌｉｎｅｐｒｏｃｅｓｓ）建立较精确的目标函数，采用并行计算方法计算出速度场，然后以此为依据实现对目标图像的并行分割。实验结果表明：该算法不仅能适应多种复杂场景，而且分割精度高、速度快。It is described in this paper that a new algorithm based upon imagesequence anlysis has been developed for segmenting the image of moving object incomplex background.That is,a coarse target function is firstly set up with aMarkov random field (MRF) model and a continuous line process in use,then amotion vector field is estimated by a parallel algorithm developed, and finally,theimage of the moving object is successfully segmented on-line. It is illustrated byexpericments that this segmentation method is worth while detecting moving objectsin some various complex backgrounds with high segmentation precision and fastcomputation speed.

关 键 词：马氏过程 图像处理 运动估计 

分 类 号：TN919.8[电子电信—通信与信息系统]