Research On Motion Target Detection And Tracking Technology Based On Autonomous Robot

The essence of autonomous robot for moving object detection and tracking is to enable the robot to have the ability of moving object detection and tracking similar to human vision.The core technology of autonomous robot is moving object detection and tracking in dynamic background.Under the current dynamic background,the technology of moving object detection and tracking has the disadvantages of low detection accuracy and unstable tracking.In view of the shortcomings of existing moving object detection and tracking algorithms,this paper makes a deep research on moving object detection and tracking technology in dynamic background.Its main innovations are as follows:Aiming at the problem that traditional algorithms are difficult to detect moving objects accurately under dynamic background,a saliency analysis based moving object detection algorithm is proposed in this paper.Based on the super-pixel segmentation of single frame image,the global color difference map and spatial position difference map are established according to the difference between edge super-pixel points and central super-pixel points.At the same time,the hybrid dynamic texture saliency map is established by using sequence images,and the initial saliency map of moving objects is obtained by fusing the three saliency maps.Then,the initial saliency map is optimized by the automatic updating mechanism of cellular automata.Finally,the moving target is obtained by the local threshold segmentation algorithm.The experimental results show that the proposed algorithm has high detection accuracy.In order to meet the real-time requirements of tracking algorithm in robotic applications,and to deal with the problem of fast target movement and scale change,this paper chooses KCF kernel correlation filtering algorithm with high real-time performance as the framework of target tracking algorithm.Firstly,the LAB color histogram and HOG gradient histogram are adaptively fused to improve the target feature expression ability.Secondly,Kalman filter is used to predict the target state,which provides additional sampling points for KCF algorithm and enlarges its search range.Finally,according to the displacement relationship between adjacent frames,a scale pool is established to complete the target scale estimation,so as to improve the detection accuracy of KCF algorithm.The experimental results show that the accuracy and success rate of the proposed algorithm are 14.0% and 11.4% higher than those of KCF algorithm,respectively.Finally,based on the strategy of first detection and then tracking,the autonomous tracking of moving objects in the scene is completed,and the algorithm is validated on the NAO robot to realize the autonomous tracking of moving objects.