Research On Machining Burn Degree Of Workpiece Based On Color CCD Image

In production, machining burn degree of workpiece is usually identified by the burn color on its surface, based on empirical judgment. In order to improve the reliability of the machining burn evaluation, the burn degree of workpiece surface is researched in the paper by using of the color CCD image, where the color feature is used to judge the burn degree. The main work and the results are as follows:(1) According to the burn color feature of metal workpiece, the HSV color space for color images is selected as the research means of the machining burn degree in order to overcome the shortages of gray images in which a large number of color information will be lost.(2) Basing on the supervision color constancy algorithm for finite-dimension linear model, the effect of the light intensity on the color image obtained was eliminated. The results showed that the difference between the color images and the setting standards is decreased by 3.9%, 2.4% and 9.6% respectively for the red, green, blue monochromatic light illumination conditions. So, the foundation is laid for the feature extraction of the image color and the classification of the burn degree.(3) The feature of the color moments of the burn surface images with same surface roughness is extracted of red, green, blue monochromatic light and composite white light. The results showed that there is an obvious change in the mean hue value and mean brightness value of the image with the change of the surface burn degree of the workpiece. As the burn becomes serious, the mean hue value will increase obviously while the mean brightness is decreased.(4) Gaussian function is used to calculate the similarity between burn colors and standard colors. In combination of the percentage burn area and the burn degree expression, the whole machining burn degree of the workpiece is discriminated.(5) The grading system of the workpiece surface burn is developed by VC + + and MATLAB. Five levels of the burn samples were tested respectively, and the test result is comparatively ideal. So, the system is effective.