Abstract: Computer vision combined with digital image processing and pattern recognition techniques were evaluated for the detection of diseased tomato leaves infected with leaf mold（Fulvia fulva），early blight（Alternaria solani），late blight（Phytophthora infestans），and leaf spot（Corynespora cassiicola）. An image acquisition system was established to acquire leaf images. The image pre-processing techniques were applied to segment the lesion regions from the diseased leaves. And then nine color characteristics，five texture characteristics and four shape characteristics of the lesion regions were extracted. To classify the four kinds of tomato foliage diseases，stepwise discriminant analysis combined with Bayes discriminant analysis and principal component analysis combined with Fisher discriminant analysis were executed to develop the discriminant models. By the stepwise discriminant analysis，we selected 12 characteristics from the original 18 variables to develop the Bayes discriminant function, and results showed that the classification accuracies for the training and testing sets achieved 100% and 94.71% respectively. By principal component analysis，the 18 variables were reduced to two principal components（PCs）. The classification model based on the two PCs achieved classification accuracy of 98.32%. These results indicated that it is feasible to identify and classify tomato diseases using computer vision technology. This preliminary study, which was done in a closed room with restrictions to avoid interference of the field environment, showed that there is a potential to establish an online field application in tomato diseases detection based on computer vision and image processing techniques.

Key words: computer vision, tomato disease, feature extraction, stepwise discriminant, principal component analysis, discriminant model