摘要
静态背景下运动目标检测的抗噪性能较差。为此,提出一种改进的运动目标检测算法。对原始图像进行预处理,将五帧差分和背景差分相结合,利用基于自适应背景模型的动态阈值,提取图像中的运动区域,并进行形态学滤波和连通性检测,最终获取运动前景目标。实验结果表明,该算法能完整提取运动目标,背景适应性强,实时性好。
Aiming at moving object detection under the static background, this paper proposes a modified difference fusion algorithm. The proposed algorithm preTprocesses the original image, combines with background difference and five frame difference, improves the method of automatically extracting background, it extracts moving regions in images by means of the dynamic threshold based on the adaptive background model, and gets the final moving objects by morphological filtering and connectivity detecting. Experimental result shows that the algorithm can extract moving objects more completely, with background more adaptable and real-time guaranteed.
出处
《计算机工程》
CAS
CSCD
2012年第4期146-148,共3页
Computer Engineering
基金
上海市教委重点科研基金资助项目(06ZZ36)
关键词
帧间差分
背景差分
背景模型
动态阈值
边缘检测
运动目标检测
frame difference
background difference
background model
dynamic threshold
edge detection
moving object detection
