Estimation of Physical Characteristics of Noisy Peach Leaves Using a Unified Algorithm

Chunxian Chen; Haixin Wang

doi:10.47852/bonviewAIA52025928

Authors

Chunxian Chen Southeastern Fruit and Tree Nut Research Laboratory, USDA Agricultural Research Service, USA
Haixin Wang Department of Mathematics and Computer Science, Fort Valley State University, USA

DOI:

https://doi.org/10.47852/bonviewAIA52025928

Keywords:

bilateral filter, CIELAB color space, Gaussian noise, K-means clustering, peach leaves, physical characteristics, Poisson noise

Abstract

Peach production and breeding benefit from digital agriculture and image processing technology. However, there is little knowledge about the optimal image denoising and processing algorithms for estimating the physical attributes of peach leaves from images. The objective of this study was to evaluate different image denoising and processing techniques to obtain optimal methods and establish a unified approach for estimating the physical characteristics of healthy peach leaves, including length, width, perimeter, and area. Twenty denoising filters were evaluated, and the bilateral filter was determined as the optimal filter for peach leaf image processing with Gaussian and Poisson noise. Twenty-four color space models, including twenty vegetation indices, were evaluated, among which the CIELAB (L*a*b*) color space performed best in the segmentation task using K-means clustering. Seven segmentation algorithms were evaluated and K-means clustering was found to be the optimal method based on a custom metric function. The unified algorithm including optimal denoising, color space selection, and segmentation techniques successfully estimated the physical characteristics of peach tree leaves. The results show that a unified approach is reliable and accurate for estimating the physical characteristics of peach leaves from images. These techniques are crucial for efficient orchard management and other digitalization-based applications in peach production and breeding.

Received: 14 April 2025 | Revised: 15 September 2025 | Accepted: 28 September 2025

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

The data that support the findings of this study are openly available in the PlantVillage Dataset on Kaggle at https://www.kaggle.com/datasets/abdallahalidev/plantvillage-dataset.

Author Contribution Statement

Chunxian Chen: Conceptualization, Methodology, Validation, Investigation, Resources, Data curation, Writing – review & editing, Supervision, Project administration, Funding acquisition. Haixin Wang: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Writing – review & editing, Visualization, Supervision, Project administration, Funding acquisition.