A Comprehensive Review of Clustering Techniques in Leaf Image Processing for Plant Analysis

Abstract

Applications including disease diagnosis, species identification, and phenotypic trait evaluation are made possible by leaf image processing, which is crucial to automated plant analysis. Clustering algorithms are one of the most popular image analysis approaches for grouping visually related regions in leaf images without the need for annotated data. This makes them appropriate for agricultural settings where manual annotation is difficult. The clustering methods used in leaf image processing for plant analysis are thoroughly examined in this article. Partitional, hierarchical, density-based, fuzzy, and hybrid clustering techniques are comprehensively categorized in the paper, and their efficacy in tasks like leaf segmentation, lesion localization, and feature grouping is discussed. To provide a cohesive analytical framework, popular preprocessing procedures, feature extraction techniques, and clustering evaluation metrics are also examined. Additionally emphasized are recent developments that combine clustering with machine learning and deep learning models, highlighting their capacity to tackle issues with illumination variance, backdrop complexity, and leaf morphological diversity. Lastly, this review highlights important research issues and suggests future areas of inquiry to strengthen the reliability and effectiveness of clustering-based leaf image analysis systems.