摘要
针对帧间差分检测运动区域抗噪性差、某些部位无法完全恢复、所提取的运动目标容易产生空洞的问题,提出一种基于水平集分割的3DOGHM运动目标检测算法,在3DOGHM分离运动区域及背景的基础上,采用一种改进的水平集进化模型进行运动目标分割。实验结果表明,该算法抗干扰能力强,可以更准确、完整地检测出运动目标。
According to the fact that motion region detected by commonly adopted frame difference has bad noise resistance ability, and cavity exits, this paper proposes the method of 3D Orthogonal Gassian-Hermite Moments（3DOGHM） for detecting moving objects based on level set segementation. This method uses 3DOGHM separate motion region and background, and adopts an improved level set segmentation motion object. Experimental results show that this algorithm has strong anti-interference ability and can detect motion object much completely and the problem of existing cavity is improved greatly.
出处
《计算机工程》
CAS
CSCD
北大核心
2010年第7期230-232,共3页
Computer Engineering
基金
国家科技支撑计划基金资助项目(2007BAG06B06)
关键词
三维高斯赫密特矩
运动目标分割
水平集分割
3D Orthogonal Gassian-Hermite Moments（3DOGHM）
motion object segmentation
level set segmentation
