Automated computer vision morphometry of Siberian elm (Ulmus pumila L.) leaves for the assessment of fluctuating asymmetry

Developmental stability is a sensitive indicator of environmental quality, and fluctuating asymmetry of bilateral morphological traits offers a straightforward means of its assessment. In plants, leaf blades are convenient structures for such analyses, yet traditional manual morphometry is laborious, subjective, and prone to interoperator variability. This study presents a fully automated computer vision pipeline to measure key morphological features of Siberian elm leaves (Ulmus pumila L.) from scanned herbarium specimens. The workflow comprises a custom calibration template, contour detection, morphological filtering, row-wise sorting, and automated petiole removal. Algorithms are provided for the determination of leaf area, length, width, and, critically, for fluctuating asymmetry studies, the width of the left and right leaf halves. Validation against two conventional methods, pallet counting and a geometric formula, on a set of ten leaves demonstrates that digital area measurements are consistently more precise, avoiding the systematic overestimation inherent in manual partial-square counting and the rigidity of a fixed shape correction factor. The entire pipeline is implemented in Python with OpenCV and delivers metric-scaled results in tabular form, drastically reducing operator effort while ensuring full reproducibility. The proposed approach is suitable for large-scale environmental monitoring and can be embedded in a user-friendly software application.