A bio-inspired framework for apple leaf disease detection: Integrating lesion localization, ant colony optimization, and machine learning
Apple trees, among the most widely cultivated and economically important orchard species, are highly susceptible to foliar diseases such as Black Spot, Black Rot, and Cedar Rust. Due to the visual similarity of symptoms, accurately distinguishing among these diseases poses a major challenge. Convent...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-12-01
|
| Series: | Smart Agricultural Technology |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2772375525003739 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Apple trees, among the most widely cultivated and economically important orchard species, are highly susceptible to foliar diseases such as Black Spot, Black Rot, and Cedar Rust. Due to the visual similarity of symptoms, accurately distinguishing among these diseases poses a major challenge. Conventional diagnostic approaches, such as expert visual assessments and laboratory analyses, are often time-consuming, costly, and limited to post-symptomatic stages. To address the growing need for rapid, accurate, and scalable solutions in precision disease detection and management, this study presents a novel framework integrating image processing, artificial intelligence (AI), and ant colony optimization (ACO) for automated disease classification in apple leaves. The proposed method comprises five key steps: (1) background removal from leaf images, (2) diseased area detection, (3) extraction of texture, color, and shape features, (4) feature selection using ACO to identify the most informative attributes, and (5) disease classification using a support vector machine (SVM) classifier. Experimental results demonstrate that preprocessing steps, particularly background removal and lesion localization, significantly enhance classification accuracy. The system achieved class-wise accuracies of 95.12 % (Black Spot), 90.91 % (Black Rot), 94.87 % (Cedar Rust), and 88.89 % (Healthy), with an overall classification accuracy of 92.50 %. Among the features used, texture contributed most significantly to performance, followed by color and shape. These findings highlight the effectiveness of combining diverse image features with bio-inspired optimization techniques for plant disease detection and offer a promising direction for future research and deployment in intelligent agricultural monitoring systems. |
|---|---|
| ISSN: | 2772-3755 |