Abstract Automatic mobile ball-picking robots have a wide range of applications in sports. In this paper, the target localization recognition for ball-picking robots was studied. Taking tennis ball as an example, the image was segmented and processed based on hue, saturation and value (HSV), and the tennis balls were recognized by Canny edge detection and Hough circle detection. Finally, the tennis balls were localized based on the depth camera. The experiment found that the method designed in this paper effectively identified the tennis balls, with a high localization accuracy, and the errors on the X and Y axes were below 25?mm. The results verify the effectiveness of the designed method in recognizing and positioning tennis balls. The method can be applied in practice.
展开▼
