Table Tennes Location Application In The Robot Tracking Based On OPENCV

Robot development is rapid, industrial robots have been very mature, abb, funac, motocom and a series of industrial robots have been widely applied in some large assembly, welding and other enterprises, as China's manufacturing 2025 major state planning is put forward, make the development of robot into the rapid rise. At the same time, service robot as an important part of the state planning into people's horizons, sweeping the floor, serving robot one after another into the family, such as restaurants.Just according to the scheduled track service robots already meet the needs of daily, has the autonomous learning, decision-making service robot is the main direction of scientific research. And machine vision is the core of service robot autonomous learning, decision-making, dynamic target identification, target tracking, abnormal behavior detection actual demand is the main trend of pattern recognition, machine learning formally. However, the robot camera movement brought the scene, background modeling methods such as do not adapt to this new challenge.This paper focuses on target camera movement cases(table tennis) recognition technology. Mobile robot is widely applied to many fields, such as military, production, living and so on, it has received widespread attention around the world. The Intel computer vision processing library named OpenCV is welcomed by researchers and developers for it's openness and practicality. Becomes one of the most important open source development tools of intelligent robot visual processing system. This research uses automatic walking robot as a test platform, single camera as visual basic, deeply studies camera calibration, background modeling, threshold operation and Kalman tracking with Visual Studio 2010 and OpenCV to implements these algorithms.Firstly, According to the research content, using the camera calibration method based on circular feature to get the parameters of the vision system calibration. Then input images for pretreatment and background modeling according to the environment. To meet in the requirements of robot mobility, this research considers both robustness and real-time.Secondly, As the goal of this subject, we extracts target feature to achieve identification and positional information of table tennis in Color space. At the same time, in order to eliminate other interference such as the same color objects, the Visual signal is preprocessed by Kalman filter.Finally, the output of simulation system based on image coordinates of the center coordinate system and prospects of the pixel parameter, communicating with the center computer robot tracking function of table tennis.The simulation results show that the proposed tracking method has higher tracking precision and veracity, in different environment test showed high robustness, having a certain ability of practical application.