Revolutionizing Sustainable Agriculture with SoilFeX: A Multi-modal Soil Feature Extraction Approach

Soil health prediction is the important for the agriculture and it is directly impact on choice of the crop, productivity and environmental sustainability. Large-scale agriculture or real-time applications, conventional laboratory-based soil testing is time-consuming, labour-intensive, and costly. To focus the limitations this research proposed SoilFeX (Soil Feature Extractor). This is a holistic and multi-modal image-based algorithm focused on soil feature extraction automation through advanced image processing methods. This method is having series of components such as Multi-Scale-Directional-Gray-Level-Co-occurrence-Matrix (MSD-GLCM) for texture analysis, Multi-Scale Rotation-Invariant Local Binary-Patterns (MS-RI-LBP) for structural feature analysis, and Color-Auto-Correlogram (CAC) for spatial color assessment. It also extracts spectral features such as the Fourier Transform, Soil Moisture Index (SMI), and Normalized Difference Nitrogen Index (NDNI) to measure water content and nitrogen presence in the soil. This model also allowing holistic assessment of soil health based on regression integrates these characteristics to predict soil pH.