Research On Ferrographic Image Segmentation And Wear Particle Feature Extraction Technology

As one of the most effective solutions for machine status monitoring and fault diagnosis, ferrography technology has already played a great role in the industrial field. However, as the traditional ferrography technology could not meet the actual requirement due to shortages such as subjectivity, low precision and time consuming, it is limited for the wider application. It is urgent to resolve the problem of intelligent ferrography diagnosis. Thus, the automatic recognition technology for wear debris is a key research topic that so many researchers, whether in overseas or domestic, have paid more attention on it in last decades.In this paper, Visual C++.net 2003 is adopted as the software development platform, OpenCV library as an auxiliary tool for processing on ferrograhic image. In order to achieve the goal of ferrography technology intelligentization, the current techniques for wear particle segmentation and feature extraction have been given more in-depth study. By using computer image processing technology, the ferrographic image smoothing, filtering and morphology processing have been carried out. It has achieved effective segmentation to the ferrographic image with adhesion particle by using the improved watershed algorithm. To get comparatively accurate segment for all particles of different colors, the color segmentation algorithm based on dynamic threshold has been proposed in this paper. This paper also gives more in-depth study in the field of particle feature extraction which lay the foundation for pattern recognition.Although the internal combustion engine is selected as the test object related to this research, but the research and the method in this paper can be used for other machines and equipments. It is of great development and wide application of the intelligent oil-monitoring technology.