Weleing Image Processing Based On Adjustable Laser Vision Sensor

In the actual work environment, due to welding working conditions can not be very satisfactory;and the arc sensor is near to the weld acr ,there are arc strong, high temperature and radiation, electromagnetic interference, dust flying and other negative factors.Therefore, the images acquired by sensor can not be the ideal image, it needs some processing steps to improve the image quality for the observation and evaluation, extract useful information parameters of the control system information parameters [1-4].Therefore, welding image processing is very important to realize real-time seam tracking.But in China, with independent intellectual property rights and being able to truly applied to the actual production process based on the image sensor of the seam tracking system is also very few.In terms of its development trend, with independent intellectual property rights, reliable performance, low-cost image processing system will be the main future direction of development.In this paper, a weld seam image real-time acquisition and processing system based on adjustable laser vision sensor is developed.The system uses a wavelength of 650nm of the AlGaInP semiconductor laser as a light source, with an area-array CCD camera to receive the V-shaped weld groove strip light reflecting from workpiece , using the NI PCI-8254R acquisition card Laser to acquire the laser strip image, the image processing and weld feature points acquired are processed in the Labview assistant . Laser sensor is made of mainly by the semiconductor, industrial CCD, 650nm narrow-band filter mirror and so on.Through the experiment, getted the installation position and structural parameters of the vision sensors which ensure to get the exact parameters of the image in the seam extraction.Welding process will produce a strong arc, splash , dust interference , and this seriously affected the gain of the weld seam feature points .In this paper, a series of weld image pre-processing are used to reduce such interference, including those based on histogram revised image binarization, mathematical morphological closing operation, based on clustering algorithm removing isolated points.After the pre-processing steps ,through the Hough transform to get straight lines of the weld seam, and then seek a straight line features of the intersections which will be show the weld seam feature points.Finally, processing a large number of real-time collected images, to get the actual weld deviation,it can see from image processing results ,the difference between the computerde debiation and the actual deviation of the between is very small which indicats the image processing effect is very good.