Analysis Of Wear Particle Characteristics Based On Digital Image

The detection of mechanical wear conditions has always been an important topic in industrial and academic research,and the wear condition is closely related to the abrasive particles generated by wear,so studying the characteristics of abrasive grains is a very good means of studying mechanical wear.The characteristic analysis of abrasive grains is a method of indirectly obtaining the wear condition by analyzing the abrasive grains,and the running condition of the machine is known from the wear condition.In view of the deficiencies of manual analysis of abrasive particles,based on the domestic and international analysis methods of abrasive grain characteristics,this paper proposes a digital image based image based on digital image processing and pattern recognition algorithm combined with the characteristics of wear particles in tribology.The analysis method of abrasive grain characteristics was carried out for the shape and texture of abrasive grains,and was realized by using related programs.In this paper,the electron microscope and its supporting drive are used to complete the collection of the abrasive image.In the Visual Studio 2015 development platform,the C++ and OpenCV functions are used to preprocess the image,divide the background,grind the count,and outline.extract.Then the shape feature is extracted from the image,and the radius of curvature and some geometric parameters are extracted.The texture features are extracted from the abrasive grain.The texture feature is composed of the detail texture feature and the global texture feature.The extraction method is: detecting the Harris corner.The LBP operator is used to extract features from corner points.The extracted features are clustered using K-Mean to form detailed texture features.LBP feature extraction is performed on the whole image,and histogram statistics are performed to form global texture features.Finally,the BP neural network was used to classify the shape of the abrasive grains.The KD tree algorithm was used to classify the abrasive grain texture.The recognition rate is similar to the artificial identification,which meets the actual engineering requirements.The method of detecting and analyzing the abrasive grain characteristics based ondigital image is obviously superior to the conventional method of artificial experience detection,and realizes the functions of image acquisition,preprocessing,recognition,wear analysis,etc.It has the advantages of high efficiency,accurate identification and high reliability.