Superlative Approach for Plant Disease Identification with Enhanced CSA Algorithm

Abstract

Disease detection in plant leaf helps farmers to protect the plant from diseases at its early stage. The most important problems are determining and anticipating plant diseases, which may be addressed for increasing output. In this research, Rider Cuckoo Search algorithm is improved with K nearest neighbour algorithm is used to classify the diseased leaf. Initially the Gaussian filtering is used in pre-processing to remove the noises in image. Following getting pre-processed image, it is exposed to segmentation step, which uses piecewise fuzzy C-means (piFCM) clustering to acquire the segments. Segmentation involves the feature extraction process which has information gain, histogram of oriented gradients (HOG), and entropy. Finally plant Disease is classified using the KNN algorithm. This proposed algorithm is implemented with the images of the plant village dataset. The proposed research work is evaluated using certain parameters like accuracy of the disease detection, Error of the algorithm, Speed of the algorithm, and time for classifying the disease. The Proposed algorithm outperformed with the values of 99.32% accuracy, 0.68% error, 2400 obs/sec speed, and time taken is 0.57743 sec respectively when compared with the existing algorithms like Hybrid SIFT algorithm, Hybrid K-means Fuzzy logic SVM algorithm, and Cuckoo Search Algorithm.