Study On Double-Reflector Laser Sensor

Weld seam tracking technology based on vision sensor has attracted the interest of researchers, because it has access to rich and large information, high stability, and wide use, and so on. Combination with the latest research result of computer vision and image processing, Seam tracking technology based on vision sensor will greatly increase welding robot's abilities to adapt to the outside and is expected to be one of the most promising seam tracking sensors. But the price of laser sensor abroad is expensive and the real commercial products about laser sensor which are made in China are not on sale, despite of many domestic studies. So it is very important to study in-depth laser vision sensing technology and develop laser structure light vision system with independent intellectual property right.In this paper, laser vision-based seam detecting system, named seam tracking system based on double-reflector laser senor, which is composed of laser light, filtering system and imaging system, is developed. The source of the system is determined through researching the interference light (Arc light) characteristics and the characteristics of the laser light source and clear welding images are obtained. As to pin-hole model with radial aberration, the traditional two-step camera calibration is improved and a plane calibration method is designed, which has obtained the exterior and inner parameters of the vision system. And this paper develops an algorithm of linear least squares fitting to extract the single-pixel line of calibration grid and obtain the coordination of calibration feature points by crossing the lines. The seam tracking principle of this vision system is described.Measurement algorithm of V-groove weld seam is developed. The relationship between the actual location deviation and the weld spot's deviation in the weld seam digital image.According to the traits of weld image, algorithm library is built to process image to detect the weld seam. To get a clear weld image, iterative threshold method is developed to process the image binary. The framework of weld seam is put forward by use of skeleton and thinning method in mathematical morphology. The feature points of V-groove weld seam is found through the relation of straight line's slope. The test based on V-groove weld seam is took and good result is got.