Interpretable Leaf Vein Analysis Using Classical Machine Vision

Developed an interpretable computer vision pipeline for analyzing leaf vein structures from digital images, focusing on classical machine vision rather than black-box deep learning. The project explored how image processing and feature extraction techniques can transform raw leaf imagery into meaningful structural information for quantitative and visual analysis.

Key Contributions & Technologies:

● Designed an end-to-end image processing pipeline for extracting and analyzing leaf vein structures.
● Applied preprocessing, enhancement, filtering, and segmentation techniques to improve vein visibility and reduce image noise.
● Used morphological image processing to refine extracted vein structures and improve structural representation.
● Developed feature-extraction workflows to convert visual vein characteristics into measurable information.
● Maintained interpretability throughout the pipeline by visualizing intermediate processing stages and resulting structural features.

Research & Development Approach:

● Used a classical machine vision approach emphasizing transparent and explainable image transformations.
● Iteratively refined preprocessing, segmentation, and feature-extraction stages to improve robustness across leaf images.
● Evaluated processing results visually and quantitatively to understand the impact of individual pipeline stages.

Tools & Technologies:

● Python & OpenCV
● Classical Computer Vision
● Image Processing & Segmentation
● Morphological Operations
● Feature Extraction & Quantitative Analysis
● NumPy & Visualization

Outcome:

Developed an interpretable machine vision workflow for extracting and analyzing leaf vein structures, providing a foundation for applications in plant phenotyping, biological image analysis, feature-based classification, and future machine learning integration.