一种面向停车场场景的运动目标检测与跟踪方法

Park Scene Oriented Moving Objects Detecting and Tracking Method

针对采用固定相机的停车场场景监控视频中可能出现运动目标较长时间停留的情况,提出一种帧间差(Frame difference)和运行期均值(Running average)相结合的运动目标检测方法,然后在卡尔曼滤波以及运动目标直方图和轮廓信息的辅助下实现运动目标的跟踪.实验结果表明,该方法可以在满足实时监控的需求下较好的检测与跟踪停车场场景中的运动目标.

To solve the problem that moving objects may be still for a long time in park scene＇s surveillance video captured by static camera, this paper proposes a moving object detection method containing both Frame difference and Running average, then designs tracking method using kalman and moving object＇s histogram and contour information. The experiment results show that the proposed method can preferably detect and track moving objects in park scene and meet the need of real time.