3D Reconstruction of Plants Under Outdoor Conditions Using Image-Based Computer Vision

Abstract

3D reconstruction of plants under outdoor conditions is a challenging task, for applications such as plant phenotyping which needs non-invasive methods. With the availability of new sensors and reconstructions techniques, 3D reconstruction is improving rapidly. However, sensors are still expensive for researchers. In this paper, we propose a cost-effective image-based 3D reconstruction approach which can be achieved by off-the-shelf cameras. This approach is based on the structure-from-motion method. We implemented this approach in MATLAB and Meshlab is used for further processing to achieve an exact 3D model. We also investigated the effect of different adverse outdoor scenarios which affect quality of 3D model such as movement of plants because of strong wind, drastic change in light condition while capturing the images. We have decreased the appropriate number of images needed to get precise 3D model. This method gives accurate results and it is a fast platform for non-invasive plant phenotyping.