A Pilot Image Analysis Pipeline for Automated Leaf Morphology in Pithecellobium dulce

Introduction: Leaf morphology is vital for plant identification, but traditional methods are subjective and inconsistent. Methods: This pilot study presents an image analysis pipeline for Pithecellobium dulce leaves using ImageJ and MATLAB. Steps included grayscale conversion, Sobel/Canny edge detection, GLCM texture analysis, and SSIM comparison. Results: Canny edge detection showed higher edge density than Sobel. Texture metrics were consistent, and SSIM scores (0.6700–0.699) indicated high structural similarity among leaves. Discussion: Canny edge detection captured finer venation than Sobel, while GLCM and SSIM confirmed strong structural similarity among leaves. The pipeline demonstrated reproducible, objective, and scalable quantification of leaf morphology, reducing observer bias and enabling automated phenotyping. Conclusion: The pipeline offers reproducible, objective leaf analysis, reducing bias and supporting applications in taxonomy and digital phenotyping.