Detection of Pepper Leaf Diseases Through Image Analysis Using Radial Basis Function Neural Networks

Abstract

Pepper (Piper nigrum L.) is a high-value cash crop and plays a significant role in Indonesia's agricultural sector. However, pepper production is often hindered by diseases that affect the plant's leaves. This study aims to develop a pepper leaf disease detection model based on image analysis using a Radial Basis Function Neural Network (RBFNN). Conventional methods relying on expert visual assessment are often inefficient, especially on a large scale. In this research, image preprocessing was performed by transforming the images into the CIELAB color space and using K-Means Clustering for feature extraction. Texture feature extraction using the Gray Level Co-occurrence Matrix (GLCM) provides rich information about patterns and intensity distribution in the images, which is effective for distinguishing disease classes. The RBFNN algorithm is then used to identify diseases by capturing the complex non-linearities in the data. Based on the testing results, this model achieved an accuracy rate of 91.67%, demonstrating excellent performance.