Application Of Parallel Processing In The Bearing Surface Flaws Inspection

It is very important to bearing quality detection that the surface flaws inspection. It directly relate to the length of bearing life which is mounted. At the present time, the method which we used is check in sight. It not only waste a lot of time, but also make the uninspected bearings easily neglect.In this paper, design a system to automatically on-line bearing inspection based on CCD camera, digital image processing, and control technology of PLC. The system includes image collection, pro-process, image character extraction, it can automatically differentiate and pick-up available information of the original image by image processing and segmentation; may automatically detect the area size of flaws by edge detection. These nuclear algorithms of image processing adopted that MMX/SSE optimized, ensure real-time process these images. We also use parallel connection among three CCD cameras and three pieces of image collection clip, then the data stream transmit PC by the PCI bus, ensure real-time collection. These algorithms of parallel processing were adopted. This can well resolves the problem between real-time process and the accuracy of inspection, So this can reduces the load of worker and raising productivity.