Plant Leaf Diseases: Feature Extraction and Identification Using Convolutional Neural Networks and GLCM features

The economics of a nation is greatly impacted by agricultural production, and identifying diseases of plant leaves is essential to preserving yield quantity and quality. This study demonstrates a novel deep convolutional neural networks (DCNN) technique for detecting plant leaf disease. Plant leaf photos are first preprocessed using AMA (Adaptive Mean Adjustment) method to improve them and used a 2D Adaptive Anisotropic Diffusion Filter (hereafter 2D AADF) to remove noise. An improved method of segmentation is to use Adaptive Otsu (AO) thresholding along with Enhanced Fast Fuzzy C Means Clustering (hereafter EFFCMC). After extracting Grey Level Co-occurrence Matrix (GLCM) features from segmented pictures, Principal Component Analysis (PCA) is used to reduce the dimensionality. In experiments, the new DCNN architecture-which consists of two fully linked layers, a SoftMax layer, and four convolutional layers-achieves better classification results.