Application of Image Processing and Advanced Intelligent Computing for Determining Stage of Asian Rust in Soybean Plants

This paper presents a new method that uses the advanced techniques of digital image processing and computational intelligence for monitoring and identifying the stages of Asian rust (Phakopsora pachyrhizi) in soybean plants (Glycine max (L.) Merril). Its establishment included organization and structuring of digital images of soybean plant leaves, image preprocessing steps, and segmentation based on the phenomenology of the disease development process with a semantics approach involving agricultural evaluation. The method also required recognition of the patterns appearing on leaves due to the presence of the disease as well as machine learning using a support vector machine for the classification and interpretation of the stages and their evolutions. For the stage of pattern recognition, the techniques of feature extraction scale-invariant feature transform, Hu invariant moments, and histogram of oriented gradients were used and principal component analysis was conducted for the dimensionality reduction of the integrated vector of the features. The results of the application of the method showed potential for the monitoring and identification of the stages of the disease for determining the subsidies in the decision-making of production systems by agricultural producers.