A Hybrid Inspection Method for Surface Defect Classification

摘要

A vision-based inspection method based on rough set theory, fuzzy set and BP algorithm is presented. The rough set method is used to remove redundant features for its data analysis and procession ability. The reduced data is fuzzified to represent the feature data in a more suitable form as input data of a BP network classifier. The classifier is optimised using uniform design. By the experimental research, the hybrid method shows good classification accuracy and short running time, which are better than the results using BP network and neural network with fuzzy input.