检测运动物体需要无运动物体的背景图像，所以，首先应用多帧像素平均值法提取了运动视频序列的背景图．从背景图像中分离目标像素，获取目标的质心坐标，并应用质心跟踪法以灰色图像序列为基础，对运动的目标进行实时检测和跟踪。质心跟踪法的目标位置通过质点的中心来确定，该算法计算简单，计算量小，其稳定性与精度主要取决于序列图像的分割及其阀值的确定情况。文中给出了用Opencv实现算法的具体过程和关键代码。并且设计了跟踪运动车辆的控制界面，方便了实时监控。实验结果表明，该方法可以实现视频序列中运动目标的识别，具有实时性、并能给出较好的识别效果。%Detecting of moving object need stationary object as background image,so we should pick up the background picture of movement＇s video sequence with multi-frame pixel of average algorithm. Separating target pixels from background picture,we will get the target coordinates , and apply centroid tracking algorithm to detect and track the moving target based on gray image sequence.The target＇s position can be determined by the center of particle.This algorithm uses small amount of calculation and easy to work.Its stability and precision mainly depends on image segmentation and the value of threshold.The detailed process and the key code of Opencv algorithm had given in the article,also the system designs a control interface, so that the manager can monitor the movement of vehicles at anytime. Experimental results show that using this way can achieve identification of the moving target with real-time.
展开▼
