Analysis of depth measurement errors of tomatoes using binocular stereo vision based on single factor experiments

Abstract

To improve the measurement accuracy of depth of tomatoes based on binocular stereo vision, the fluctuation of depth measurement errors of tomatoes caused from every influence factor was analyzed quantificationally based on single factor experiments after the analysis of influence factors resulted in depth measurement errors. The depth measurement method for tomatoes based on binocular stereo vision included four steps, image segmentation, multi-fruits matching, stereo matching and depth measurement. Three stereo matching methods were analyzed, centroid-based matching, area-based matching and combination matching. The influence factors resulted in depth measurement errors were the repeatability, the measurement method for depth reference values, the size, color and shape of tomatoes, and the lighting conditions. Test results showed that the main factor resulted in depth measurement errors was the size of tomatoes. The rule of the main factor to depth measurement errors should be researched and corresponding correction models should also be created in future researches.