Mathematical Simulation of Rice Leaf Lesions for Machine Learning–Based Disease Diagnosis

Abstract.

This article presents a research based on mathematical modeling, visual simulation of leaf spots caused by plant diseases. Using analytical functions (simple Gaussian distributions, elliptical and jagged edges), we simulate the spots caused by two diseases quite common in rice crops. Geometric formulas (area, perimeter, circularity) are used for interpretation, while the shape and color features of the spots are analyzed to generate a synthetic dataset. To demonstrate the applicability of the method, we show how, for two rice diseases, adding mathematically generated images to the training images leads to increased accuracy of machine learning models for automated plant disease diagnosis. Thus, the results confirm that synthetic image generation can effectively complement real datasets, making disease detection more robust.