摘要
为提高光流法在检测追踪运动目标时的效率与准确性,提出了一种基于运动预测改进的优化光流目标跟踪算法。该算法通过光流法查询运动点,采用均值漂移聚类得出运动物体质心,通过随机选取运动目标部分角点并查询运动信息,经卡尔曼滤波预测质心运动信息,查询预测区域信息,降低算法的复杂度并提升效率和准确性。实验结果表明:在监控视频中对车辆与行人的追踪平均准确率达到了68.7%与78.5%,每秒能够处理31.5帧与34.7帧;通过对比得出,优化的光流算法有效地提高了检测追踪监控视频中运动目标的效率与准确率。
In order to improve the efficiency and accuracy of optical flow method in detecting and tracking moving targets,an improved optical flow target tracking algorithm based on improved motion prediction is proposed.The optical flow method is used to query the motion points,and the mean drift clustering is used to obtain the centroid of the moving object.Randomly selecting the corner points of the moving target and querying the motion information,the centroid motion information is predicted by the Kalman filter,and the predicted region information is queried,and efficiency and accuracy are increased.The experimental results show that the average accuracy of tracking vehicles and pedestrians in surveillance video reaches 68.7%and 78.5%,and it can process 31.5 frames and 34.7 frames per second.Compared with the other two algorithms,the optimized optical flow algorithm effectively improves the efficiency and accuracy of detecting and tracking moving targets in the video.
作者
高屾
朱成杰
GAO Shen;ZHU Chen-jie(Anhui University of Science and Technology,Huainan 232001 China)
出处
《新余学院学报》
2020年第2期19-24,共6页
Journal of Xinyu University
基金
安徽省光电感测工程技术研究中心开放基金项目“遥感影像地物识别与提取技术应用研究”(01001567-201601)。
关键词
目标跟踪
计算机视觉
光流法
均值漂移聚类
卡尔曼滤波
target tracking
computer vision
optical flow method
mean shift clustering
Kalman filter
