Pattern Recognition Approach for Crop Disease Identification Using Leaf Texture and Color Feature Extraction

Crop diseases reduce yield and quality, and visual inspection can be time-consuming and dependent on lighting conditions, leaf characteristics, and disease similarity. Traditional image-based recognition approaches can have redundant descriptors and confuse similar disease patterns based on visual appearance only. A novel Hybrid Color–Texture Optimization with Support Vector Machine (HCTO-SVM) pattern recognition approach is developed to identify crop diseases from complementary color and texture features. The leaf images are pre-processed and segmented; color features are extracted using the Hue, Saturation, and Value (HSV) color space, and texture features are extracted using the Gray-Level Co-occurrence Matrix (GLCM) and Local Binary Pattern (LBP). The complementary descriptors are fused, and Particle Swarm Optimization (PSO) is used for discriminative feature selection and to reduce redundancy. A Support Vector Machine (SVM) with a Radial Basis Function (RBF) kernel performs nonlinear pattern recognition for final crop disease identification. Evaluation on the Kaggle PlantVillage dataset shows 98.76% Macro-F1, 98.21% MCC, 98.14% Cohen's Kappa, and 98.63% Balanced Accuracy, demonstrating consistent recognition across the evaluated disease classes. A unified color–texture representation, optimized feature selection, and nonlinear recognition framework for robust crop disease recognition are presented. The strategy facilitates early screening, precision agriculture and scalable crop health monitoring.