Research and Development of Vision Positioning System for Evaporator Tube Plate Crawler

Considering the complexity of nuclear power plant environment and the limitation of manual check of heat transfer tube position, it is necessary to develop an efficient visual positioning system. The recognition methods of the target tube hole include: to preprocess the characteristics of the image, an improved median filter algorithm is designed, which reduces the possibility of image distortion; to obtain the contour of the tube hole, image segmentation is carried out; the identification of the tube hole based on BP neural network, which improves the parameters such as step length, reduces the vibration. On the other hand, in order to locate the heat transfer tube, based on the research of camera imaging model and calibration algorithm, Zhang Zhengyou calibration method is used to calibrate the binocular stereo vision system. Combined with the principle of 3D reconstruction, the 3D coordinates of the contour center of the target pipe hole can be obtained.