T.J. Gale (Australia)
Defect, Imaging, Neural network
This paper describes the application of artificial intelligence to a structured-light imaging system. Three approaches were considered involving classical logic, fuzzy logic and an artificial neural network (ANN). Images were taken of both faulty and non-faulty confectionery product and the effectiveness of each approach in identifying reject product was determined. It was found that the effectiveness of all three approaches was greater than 90%, with the ANN being superior. It was proposed that the ANN system be further developed to increase the capabilities of this system.
Important Links:
Go Back
