An Improved Gray Compensation Method for Surface Defect Detection of Tapered Roller

In order to improve the detection efficiency and the accuracy of tapered roller surface defects, a new method for detecting the surface defects is proposed. First, it designed a special image acquisition device to solve the problem that the roller has the property of high light. Second, it proposed an improved grayscale compensation method to cope with uneven grayscale value of the image, which was extracted by coordinated information from the origin image. At last, the defect was segmented by using the histogram clustering method. Experiments show that the improved grayscale compensation method makes the image background more balanced and improves contrasts between the defect area and the background area. Compared with Niblack and background subtraction algorithm, the result shows that this method has a better performance on time cost and accuracy.