Classification of Chili Plant Diseases Through GLCM Feature Selection and the K Parameter in the K-Nearest Neighbor

Abstract

Chili pepper (Capsicum annuum L.) is a strategic horticultural commodity in Indonesia with high economic value. However, chili plants are often infected by diseases such as Anthracnose, Fusarium Wilt, Fruit Fly, and Thrips, which can lead to significant yield losses. Early and accurate identification of these diseases is crucial for effective control measures. This study aims to classify chili plant diseases based on leaf images using the Gray Level Co-occurrence Matrix (GLCM) for feature extraction and the K-Nearest Neighbor (K-NN) algorithm for classification. A total of 736 leaf images were used, divided into four disease classes. The pre-processing stages included resizing the images to 300×300 pixels, rotation augmentation (0°, 45°, 75°, 90°), and conversion to grayscale. Textural features were extracted using GLCM at four angles, and K-NN was applied with K values of 5, 7, and 9. The highest classification accuracy of 88.19% was achieved at a GLCM angle of 0° and K=5, with an overall average accuracy across all angles of 85.06%. These findings not only reinforce previous findings on the effectiveness of GLCM and K-NN but also contribute by identifying the optimal parameter configuration (angle 0° and K=5) for the specific chili disease dataset. The results have the potential to be applied as a foundation for developing an automated plant disease detection system in the field.

Plant phenotyping relevance

葉画像から植物病害状態を推定する画像特徴抽出・分類手法が研究の中心であり、精度評価とパラメータ比較も行っているため、植物フェノタイピング手法として採用。

Code and data availability

The article describes a GLCM + K-NN chili disease classification study using 736 leaf images, but contains no dataset deposit, code release, or availability statement, and no public URLs are provided. The image dataset and MATLAB analysis are not stated as publicly available.

No evidence-backed public reproduction asset is currently recorded.