A Machine Vision System based on RGB-D Image Analysis for the Artichoke Seedling Grading Automation According to Leaf Area

Abstract

In this work, the development of a machine vision system based on RGB-D image analysis for artichoke seedling grading is described as well as its integration into a robot with the capability to handle seedlings, moving them from an unclassified plug tray to a classified one. First, the seedling RGB-D image acquisition procedure is implemented. Second, the leaf area estimation algorithm is developed, which comprises an RGB-D image segmentation algorithm and the execution of a triangulation algorithm with the points inside each region defined by the segmentation as input. Then, this area is used to assess a seedling’s quality. Third, the performance and the working conditions of the machine vision system are analyzed. Fourth, the developed system is integrated into a robotic platform that has the capability of handling and moving a seedling according to the results of the machine vision system. Finally, the results are discussed and several ways to improve the system are put forward.