Measurement Of Turning Surface Roughness Based On Image

Surface roughness means microcosmic geometry shape characteristic (error) composed of minor spaces, peaks and valleys in the machining surface. It directly influences machine and instrument's service performance and life, especially makes important sense to products with high running speed, fabrication precision and tightness request. Therefore, surface roughness measuring technique takes a very important place in engineering and technological research. Along with the development of modern industrial production and technology, people put forward higher and higher request on surface roughness measurement, thus the need of nondestructive and portable surface roughness measuring equipment emerges as the times require.This thesis compared the existing surface roughness measuring methods synthetically. Based on analyzing every method's quality, and through research of image detection and neural net technique, a new method of measuring surface roughness was put forward. Firstly, acquire texture information from turning surface's image and extract feature. Secondly, establish neural net to train and recognition eigenvalues, validate that there is a corresponding relationship between image feature and surface roughness, and thus finish qualitative measurement of surface roughness. Thirdly, extract section contour from images to proceed filtering and extract parameters. 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.This thesis adopted LED light source, camera, body microscope and computer as hardware devices, MATLAB as software platform, to build surface roughness measuring system, as well as develop the corresponding measuring software. In virtue of camera's USB high transmission speed and MATLAB's powerful algorithms library, the system realized surface roughness's measurement based on image.