The measurement of planning surface roughness by neural networks based on image

Detection and Recognition of the surface roughness in the images is a topic which has received a lot of attention in the field of image processing. In this paper, a new noncontact measurement method of surface roughness, by texture analysis, is developed based on Charge Coupled Device (CCD) image in planning operations. Firstly, the surface image of the workpieces is acquired using the A102f CCD digital camera, acquiring texture information's from planning surface's image and extracting feature. Then, establishing neural network to train, we get a corresponding relationship between image feature and surface roughness, thus finish qualitative measurement of surface roughness. Compare the above parameters with the result obtained by stylus measuring equipment, find the rule through regression analysis and create mathematical formula, which finished quantitative measurement of surface roughness. The predicted surface finish values using this measurement method are found to correlate well with the conventional stylus surface finish values.