Research On Methods For Identification Of Carbon Black And Segmentation Of Impurities In Rubber Based On Image Processing

With the rapid economic development,the consumption of rubber products focused on tire have increased dramatically,which makes the quality requirements of rubber production process become more strict.In the rubber production process,mixing is one of the most important process.The quality of mixing rubber has become a evaluation standard of rubber products.As carbon black dispersion has been an important evaluation indicator of the mixing rubber,how to accurately recognize carbon black has become a key issue of objective evaluation of carbon black dispersion.There is a low efficiency and poor accuracy problem in the traditional method for artificial identification of carbon black.The equipment with the function of automatic identification used to detect carbon black dispersion is expensive.And when the carbon black is close to rubber gray,there are some deviations,which even will result in some impurities mistakenly identified as carbon black.The above two methods has seriously affected the evaluation of carbon black dispersion.According to the above problems,This paper deeply researched in the following three aspects:image preprocessing,the selection of recognition algorithm and the removal of impurities,special works included:In the first place,For the three interference of fuzziness,noise and low contrast in the image,we proposed the methods of fuzzy function,midpoint filter and gray transform to eliminate the effects of interference on the subsequent identification of carbon black.Secondly,With the full consideration of the rubber image features,we proposed the optimization method based on the inflection point to enhance the accuracy of the subsequent identification of the carbon black.Thinking about the problems of several typical image segmentation algorithm,PCNN algorithm based on the Maximum Entropy was proposed.Through the comparison of experiments result,the identification result of carbon black by PCNN algorithm based on the Maximum Entropy was more complete and delicate.Combining the method based on inflection point,PCNN algorithm based on the Maximum Entropy solved the problem of distinguish between carbon black and rubber.Lastly,According to the morphological difference between impurity and carton black,we used area,elongation,roundness and slenderness features to dispose large particles,cut marks,pores and filamentous impurities in the rubber image.The minimum bounding rectangle obtained by MER can receive a more accurate elongation.Erosion operation solved the problem of calculating the maximum thickness of the filamentous impurities.Experiment results proved that if the region's area was larger than 613 pixels,it was regarded as large particles impurities.If the elongation larger than 6.5,it was regarded as linear impurities.If the hole's area larger than 3pixels and the roundness approaches 1,it was regarded as pores impurities.If the slenderness larger than 1.3,it was regarded as filamentous impurities.Segmentation of impurities was a supplement for identification of carbon black,it also improved the accuracy of identification of carbon black.Research on Methods for Identification of Carbon Black and Segmentation of Impurities in this paper solved the problems of low efficiency and poor accuracy caused by artificial identification methods of carbon black,further improved the accuracy of identification of automatic test equipment,and provided more accurate data base for the subsequent evaluation of carbon black dispersion.