Chili Leaf Disease Classification Using a Combination of HSV Features and Discrete Wavelet Transform with the Support Vector Machine Method

Chili leaf disease is one of the factors that can reduce plant productivity and cause losses for farmers. Manual identification of diseases often takes a long time and relies on the experience of observers. Therefore, this study aims to develop a classification system for chili leaf disease based on digital image processing using a combination of Hue, Saturation, Value (HSV) color features and Discrete Wavelet Transform (DWT) texture features with the Support Vector Machine (SVM) method. The dataset used consisted of 1,548 images of chili leaves divided into four classes, namely Anthracnose, Cercospora Leaf Spot, Fresh Leaf, and Leaf Curl Disease. Color features were extracted using mean values and standard deviations from each HSV channel, while texture features were obtained through the decomposition of Haar wavelets. Furthermore, the HSV and DWT features are combined into a single feature vector and used as input on the SVM model with the Radial Base Function (RBF) kernel. The test results showed that the proposed model was able to achieve an accuracy of 96.45%, precision of 96.52%, recall of 96.45%, and an F1-score of 96.46%. These results show that the combination of HSV and DWT features is able to represent the color and texture characteristics of chili leaves well so that it is effectively used for the classification of chili leaf diseases.

Keywords: Chili Leaf Disease, Support Vector Machine, HSV, Discrete Wavelet Transform, Image Classification