TY - JOUR
AU - Fanqiang,
AB - With the development of amusement robot, this paper is based on badminton robot, This paper propose a simple and extremely efficient tracking algorithm. Using device sensor can obtain the moving target three-dimensional coordinates, the algorithm is based on the depth of the moving target information, combining the background subtraction to remove some fixed background, the nonlinear cross filtering algorithm is proposed to remove noise, which establishes the rules for target recognition based on list sorted collection, to realize the moving target detection. According to motion model to forecast the object orientation of the next frame, combined with the a priori and real-time detection target state to the target location, at the same time by time lag compensation to further improve the real-time and accuracy. Through experiments, the results show that the prediction and actual error of the drop point are within the range of 30*30mm2, which meets the research requirements of 200*200mm2.Robot can return the ball with an 84% chance. The effectiveness of the proposed tracking and target algorithm is proved.
TI - Tracking and Locating Moving Targets Based on Kinect
JO - Journal of Physics: Conference Series
DO - 10.1088/1742-6596/1098/1/012027
DA - 2018-09-01
UR - https://www.deepdyve.com/lp/iop-publishing/tracking-and-locating-moving-targets-based-on-kinect-oaGacG7IV6
SP - 012027
VL - 1098
IS - 1
DP - DeepDyve
ER -