复杂背景下运动目标的改进实时检测算法

针对复杂背景下运动目标检测难度大、算法实时性差的问题,提出了一种改进的运动目标实时检测算法.融合背景差分、帧间差分和边缘检测的信息,划定目标区域提取完整的目标轮廓.针对图像光照突变的问题,改进了运行期均值背景更新策略,在背景更新时同步更新前景分割阈值.使用复杂背景下毛细管粘度计液位检测视频验证算法证明,边缘融合方法和实时阈值更新的背景更新算法能够克服背景复杂、光照变化、运动干扰等带来的影响,提高运动目标检测精度,实现实时检测.

Improved Real-Time Detection Method for Moving Object in Complicated Background

In order to solve the problems that moving targets detection under complicated background is difficult and the real-time performance algorithm is poor,an improved moving targets detection algorithm in real time is proposed.The information of the background difference,the frame difference and the edge detection is fused for the targets detection to extract complete targets contour by means of demarcation.In order to solve the problem of image illumination mutation,the running mean value background update strategy is improved and the foreground segmentation threshold is updated synchronously when the background is updated.The background verification algorithm of the capillary viscometer liquid level detection under complex background shows that the edge fusion method and the background update algorithm of real-time threshold update can overcome the influence of background complexity,illumination variation and motion interference,and improve the moving object detection precision and realize the real-time detection.