Optimization Of Weld Seam Image Information Based On Visual Sensing Arc Welding Robot

With the development of automation, image processing and sensor technology, modern welding technology is more and more automatic and intelligent. But it is still hot issue on how to improve welding quality, productivity, and realize automatic online monitoring in welding fields. In the automatic seam tracking system, how to obtain weld stripes images clearer, extract weld information more efficiently, and calculate seam bias more accurately is still need to be studied.In this paper, the author researched the weld images optimization of arc robot which based on structured light visual sensing. The laser sent out point light converted into a strip light by the cylindrical lens which projected onto the welding surface, CCD receives those strip light from the workpiece vertically which reflect different weld groove shape. Through acquisition and image processing, it can calculate the center position of weld strip according to the middle position of deformed fringe. Compared this position with the current torch position to obtain real weld deviation, and then give this deviation to control unit to adjust weld torch to movement to eliminate bias, In the actual welding process, due to arc, splash, dust and other interference existence, the weld images captured by CCD camera may contain a lot of noise, which will affect weld stripe information extraction.Because the previous image processing results still exists noise, breakpoint, fluctuation and low processing rate. First, the author analyzes all kinds of interference factors which will effects weld images processing results. According to the linear distribution of weld strip, though calculation the line gray accumulation distribution to determine the size of video acquisition windows, then use the image processing method, such as threshold transformation,3* 1 median filtering,the level corrosion, and thinning to process the weld images.After obtain the weld stripes, according to parameters requirements of welding deviation and the characteristics of the different groove weld, the author defined which points can be feature points. Because different type welds have different cumulative distribution of gray, the author determined the searching scope and feature points. Then, using the detection rule and tracking rule, extraction feature points and calculation weld deviation based on butt joint, lap joint, V-shaped and angle groove weld. Experimental results show that the method can extract feature points more better, extraction speed improved and better real-time. It also can provide seam more accurate data for follow-link weld tracking unit.In this paper, USB serial bus image capture card can be used for acquisition of real-time weld image. Compared with traditional data acquisition card, it can form all types of mobile image processing workstations, can work in a more suitable venue or a narrow working environment and occasions, and also provide convenient for the development of seam tracking software program.