Research Of Image Feature And Fractal On Roll Surface Morphology In Wear Process

Roller is an important part of rolling mill, the surface morphology of roll is directlyrelated to the strip products’surface quality and even the texture component. This papertakes the roll image as the research object, collecting the image information of a grindingroller cycle under different stages. Based on the MATLAB toll, the paper also study thevariation of image features and fractal characteristic in the wear process of roll. Theresearch work includes：(1) The usual processing method of image was researched, analyzed and collatedbased on roll surface image in this paper. The processing method adapted to roll surfaceimage was discovered. The foundation for realizing the image characteristic extractionwas laid.(2) The characteristic of roll surface image was extracted and analyzed. Thevariation regularity of characteristic parameter with rool wear was obtained. The resultsshow that the image features，such as circularity, equal-area-cirde radius, texture entropyand second-order moment have a regular change and can be used as quantitativeparameters to describe the surface morphology of roll.(3) The fractal theory was introduced to roll surface image analysis. The papercalculates the surface fractal dimension D by using the box counting method, andanalyzes the relationship between the fractal dimension and roll wear. The results showthat fractal dimension can be used to judge roll wear condition.(4) BP neural network model used to recognize the roll wear condition wasestablished, and the software program of the integration of image proeessing andanalysing was developed. The results show that the neural network and the software havegood applicability and reliability.All of this research work is important for warranting the surface quality of theproducts and increasing the rolling productivity. It supplys a new approach to establishthe quantitative evaluation system for the wear morphology of roll.