A Hybrid Approach for Improving Image Segmentation: Application to Phenotyping of Wheat Leaves

• DOI: 10.1371/journal.pone.0168496
• Publication Date: 2016-12-19T19:00:28Z
• Authors: Joshua Chopin; Hamid Laga; Stanley J. Miklavcic
• Keywords: Segmentation-based object categorization

In this article we propose a novel tool that takes an initial segmented image and returns more accurate segmentation accurately captures sharp features such as leaf tips, twists axils. Our algorithm utilizes basic a-priori information about the shape of plant leaves local orientations to fit active contour models important have been missed during segmentation. We compare performance our approach with three state-of-the-art techniques, using error metrics. The results show tips are detected roughly one half original error, accuracy is almost always improved than breakages corrected.