基于速度特征矢量提取运动目标的图像分割方法  

An Image Segmentation Method Based on Velocity Feature Vector for Moving Target Extraction

作　　者：李冬霞[1] 曾禹村[1] 

机构地区：[1]北京理工大学电子工程系,北京100081

出　　处：《北京理工大学学报》2000年第3期347-351,共5页Transactions of Beijing Institute of Technology

摘　　要：研究基于速度特征矢量提取运动目标的图像分割方法 .根据目标图像像素移动的一致性 ,在序列图像中利用块匹配法进行帧间图像配准 ,得到目标图像块的速度估计 ,将具有相同速度矢量的目标图像块聚类 ,即可分割出运动目标 .仿真实验结果表明 ,该方法能有效地对复杂背景下的运动目标图像进行分割 ,并具有较好的抗噪能力 .An image segmentation method based on the velocity feature vector of moving target was proposed. According to moving vector sameness of the pixels in a target image block, the velocity vector was estimated by inter frame image matching in an image sequence. Block matching algorithms were utilized in the matching process. All image blocks that have the same motion vectors were clustered and a moving target was identified. It was proved that the method is effective and feasible to segment the moving target from complicated background and has the better anti noise ability.

关 键 词：运动目标检测 图像分割 块匹配 速度特征矢量 

分 类 号：TN957.53[电子电信—信号与信息处理]