基于改进分水岭和形态学的视频运动目标检测  

Moving object detection based on marked watershed and morphology

作　　者：李木文 高会民 谢小英 徐漫涛[3] 

机构地区：[1]国网冀北电力有限公司秦皇岛供电公司输电运检室,河北秦皇岛066000 [2]国网冀北电力有限公司秦皇岛供电公司发展策划处,河北秦皇岛066000 [3]上海电机学院电气学院,上海200240

出　　处：《激光杂志》2014年第11期30-33,共4页Laser Journal

摘　　要：为了获得理想的视频运动目标检测结果,本文提出了一种标记分水岭和形态学相融全的视频运动目标检测算法。首先采用各向异性扩散模型对视频图像进行预处理,消除噪声对目标检测的干扰,然后差分算法提取运动目标的轮廓,并对提取的目标轮廓进行形态学操作,最后采用标记分水岭算法对运动目标进行标记和分割,并采用仿真实验测试算法的性能。仿真结果表明,本文算法较好防止了"过分割"缺陷的出现,可以准确从复杂背景检测到运动目标,不仅提高了视频运动目标的检测正确率,并且加快了视频的检测的速度。In order to obtain ideal detection results of moving object,a new video moving object detection algorithm based on labeling watershed and detection algorithm and morphology is proposed in this paper.Firstly,the image is pre-processed by diffusion model to eliminate noise interference on object detection,and the difference algorithm is used to extract outline of moving object,secondly,morphological operations is use to process the target outline,finally,marked watershed algorithm is used to segment moving object and performance is tested by simulation experiments.The simulation results show that the proposed algorithm can prevent over segmentation problems and accurately detect moving object from complex background,not only improve the detection accuracy and accelerate the speed of moving object detection.

关 键 词：运动目标检测 标记分水岭算法 形态学操作 各向异性扩散模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]