Design of Optimal Multilevel Thresholding based Segmentation with AlexNet Model for Plant Leaf Disease Diagnosis

Accurate and earlier detection of plant leaf diseases is important for precision agriculture, and also to avoid financial loss and improve crop productivity. Many of the plant leaf diseases exhibit visible symptoms, it is hard to manually identify the plant leaf diseases. Therefore, image processing based computer aided tools is essential for prompt and accurate plant disease detection. This study presents an optimal segmentation with Alexnet based feature extraction for plant leaf disease diagnosis (OSAFEM-PLDD) technique. The proposed OSAFEM-PLDD technique intends to determine and classify various kinds of plant diseases using leaf images. The OSAFEM-PLDD technique involves fuzzy filtering (FF) based pre-processing to remove the noise and improve the image quality. In addition, chicken swarm optimization (CSO) with Kapur's thresholding based segmentation approach was employed for the detection of affected leaf regions. Moreover, AlexNet method was utilized for deriving a helpful group of feature vectors and finally, functional link neural network (FLNN) based classification model is used to perform effective plant disease diagnosis. To showcase the enhanced performance of the OSAFEM-PLDD technique, a wide range of simulations are carried out on benchmark plant leaf disease datasets and the experimental outcomes highlighted the enhanced performance of the OSAFEM-PLDD technique over the recent methods.