Image Processing of Robotic Vision Sensing for Typical Joints

摘要

According to the domestic situation that the existing image processing systems of arc welding robot vision sensing,with poor generality and robustness,cannot meet the need of seam tracking in actual welding process,the hardware system of robot vision sensing for typical joints was established based on Panasonic-TA1400 arc welding robot with six-degree of freedom.The general image processing system for vision sensing was developed by using LabVIEW.The two-dimensional wavelet transform was adopted to images of the typical weld joints for noise reduction and edge detection,both effectively filtering out background noise,and almost completely retaining the edge of the laser light band.For the key technology in laser stripe extraction and characteristic point detection,simple and effective methods including the fast search row(column) one by one and the reverse quickly finding array maximum were proposed,which pay attention to the precision and speed requirements.The results show that the image processing technology can get good effect and meet the real-time demands of the tracking system.