RESEARCH ON AUTOMATIC DETECTION SYSTEM DESIGN OF ROLLING BEARING'S SURFACE DEFECTS BASED ON NEURAL NETWORK

摘要

The paper describes rolling bearing's surface defects images definition and the classification research by the connected domain's geometry features and moment invariants features. It applies BP neural network technology to surface quality analysis processing. And it also compare Minimum distance classification method to BP neural network method. Result proved that BP neural network method have a better discrimination. Finally, it carries on the discussion to the bearing surface defects images recognition system software design technology.