Application Research Of NAO Robot Target Detection And Tracking

Moving object detection and tracking technology is an important research topic in the field of computer vision,which has been widely used in security,traffic management,robot vision.It has become an important way for mobile robot platform to perceive the environment to acquire video and image information by carrying vision sensor.This paper studies the application of NAO robot moving target detection and tracking based on monocular vision.Due to the complexity and changeability of practical application scenarios,the existing algorithms cannot meet the needs of practical application.Aiming at the problems encountered in the application of classical Camshift algorithm to robot target tracking,this paper analyzes the target detection and target tracking respectively and proposes an optimization scheme.The specific work contents are as follows:1.The advantages and disadvantages of several commonly used target detection algorithms were analyzed and compared.Aiming at the problems of large amount of interference in target detection results and incomplete target extraction,a target detection algorithm was proposed by using the multi-feature fusion of target color,edge feature and moving foreground.The histogram equalization method is used to highlight the target region and reduce the influence of illumination change.The region is divided by the edge features to obtain the complete outline of the target,and the interference of holes is eliminated by combining the motion foreground information to improve the quality of target detection.The experimental results show that the fusion target detection algorithm can filter out most of the interference background,and the detected target model is relatively complete,which can provide a good support for the subsequent tracking process.2.Study the advantages and disadvantages of emergent Camshift tracking algorithm,and Kalman filtering is proposed combined with the feature of the ORB matching improved Camshift algorithm applied to the NAO robot moving target tracking,aiming at color interference tracking failures in the tracking process,using the ORB algorithm to extract image feature points and the current target template feature point matching,statistical number,the optimal match between the two images when the optimal matching number is greater than the threshold value matching success,to regain the target in the search box in the image position update algorithm.Aiming at the problem that the target is blocked by obstacles and causes target loss,the Kalman filter is used as the prediction mechanism of target motion state,and the target position parameters are updated according to the predicted position.Aiming at the problem that the monocular camera cannot obtain the distance information,the ultrasonic sensor is used to detect the obstacle information to avoid the collision between the robot and the obstacle.The experimental results show that the ORB algorithm and Kalman filter can redetermine the position of the moving target and effectively solve the problem of tracking failure caused by similar color background interference and target occlusion.