Research on recognition and localization of workpieces on rotating body surfaces based on vision recognition algorithm

Abstract Due to the rotary workpiece on the transmission mechanism in the welding automatic production line entering the next process from a certain process, a certain rotation offset will occur along the axis direction, resulting in inaccurate positioning, which in turn reduces the surface machining accuracy of the rotary body. Hence, this paper proposes a positioning method based on visual recognition. Firstly, the industrial monocular array camera is used to collect the surface image of the rotating body, and the image feature information is identified and extracted to calibrate the camera. Then, based on the shape-based template matching method, the processed image information is compared and corrected with the required position information of the workpiece, and the judgment is made in combination with the surface area feature information of the workpiece. Finally, the deviation information is transmitted to the control mechanism for correction. For workpieces of 32 mm, 45 mm, 78 mm, and 96 mm. The experimental results show that the positioning accuracy can be achieved within 20 degrees of rotation of the workpiece along the axis, and the maximum offset error is only 0.5 degrees.