A VPRS and NN Method for Wood Veneer Surface Inspection

Variable precision rough sets (VPRS) is used to reduce the redundant features in terms of its ability of knowledge reducts. An improved network algorithm including additional momentum, self-adaptive learning rate and dynamic error segmenting is presented to solve the shortcomings of traditional BP neural network (NN). The reduced features after VPRS are fed into the improved neural network proposed to inspect the defects of surface for wood veneer, which results in short training time and a high classification accuracy with a typical application in defect inspection of wood veneer.