AN EXPERIMENTAL EVALUATION OF IMAGE FILTERING ALGORITHMS FOR EARLY DETECTION OF BANANA LEAF DISEASES

Abstract

This research presents a thorough examination and comparison of several image filtering algorithms for the diagnosis and identification of diseases affecting banana leaves. For efficient agricultural management and to reduce crop losses, diseases in banana crops must be promptly identified. The visibility of disease symptoms on banana leaves is greatly improved by image processing techniques, especially filtering algorithms. This study assesses how well a number of popular image filtering techniques—such as the Canny, Sobel, Laplacian, Gaussian, and bilateral filters—perform in enhancing the quality of images of diseased banana leaves. Each algorithm's efficacy is evaluated according to important standards such edge retention, noise reduction, and computational efficiency, with an emphasis on the practical difficulties presented by field-captured images. The findings of the experiments show the advantages and disadvantages of each filtering technique, offering information on how well they can diagnose different diseases of banana leaves. The results highlight how crucial it is to choose the right image filtering methods in order to guarantee precise disease diagnosis and provide suggestions for further study in the area of agricultural image analysis.