Weed-plant discrimination by machine vision and artificial neural network

A machine vision system using a charge coupled device camera for the weed detection in a radish farm was developed. Shape features were analysed with the binary images obtained from colour images of radish and weeds. Aspect, elongation and perimeter to broadness were selected as significant variables for discriminant models using the STEPDISC option. The selected variables were used in the DISCRIM procedure to compute a discriminant function for classifying images into one of the two classes. Using the discriminant analysis, the successful recognition rate was 92% for radish and 98% for weeds. To recognise radish and weeds more effectively than the discriminant analysis, an artificial neural network (ANN) was used. The neural network model distinguished the radish from the weeds with 100%. The performance of the neural networks was improved to prevent overfitting and to generalise well using a regularisation method. The successful recognition rate in the farms was 93.3% for radish and 93.8% for weeds. As a whole, the machine vision system using the charge coupled device camera with the ANN was useful to detect weeds in the radish farms. (C) 2002 Silsoe Research Institute. Published by Elsevier Science Ltd. All rights reserved.