Inferring Agronomical Insights for Wheat Canopy Using Image-Based Curve Fit K-Means Segmentation Algorithm and...

Area of Science:

• Plant science
• Agronomy
• Computational biology

Background:

• Phenomics and chlorophyll fluorescence are key to understanding plant stress responses.
• Image-based analysis of wheat canopy morphology provides insights into plant health.
• Chlorophyll fluorescence signals indicate photosynthetic activity and stress levels.

Purpose of the Study:

• To develop and validate an algorithm for analyzing image-based morphological changes in wheat canopies.
• To identify water stress indicators in wheat using texture features derived from canopy images.
• To compare the performance of different image segmentation algorithms for wheat canopy analysis.

Main Methods:

• A three-stage algorithm involving dynamic thresholding via curve fitting, iterative K-means segmentation (CfitK-means), and computation of 23 GLCM texture features.
• Statistical analyses including correlation, factor, and agglomerative clustering were employed.
• A public repository of wheat canopy images with normal and water-stressed chlorophyll fluorescence data was utilized.

Main Results:

• The CfitK-means algorithm achieved a high IoU score of 95.75%, outperforming seven other segmentation algorithms.
• All 23 computed GLCM texture features were found to be effective in studying water stress-induced changes in wheat canopy shape and structure.
• The analyses successfully identified indicators of water stress.

Conclusions:

• The CfitK-means algorithm is a robust method for segmenting wheat canopy images, crucial for phenomic analysis.
• GLCM texture features derived from segmented images are valuable for detecting and understanding water stress in wheat.
• This approach offers significant potential for improving crop monitoring and management strategies.