TREE DROUGHT STRESS DETECTION BASED ON 3D MODELLING

Abstract

Abstract. Precise and detailed reconstruction of 3D plant models is an important goal in computer vision. Based on these models, important parameters can be extracted, which would be very useful for monitoring the tree health situation. This paper has firstly constructed the 3D plant model based on MC-CNN using close-range photogrammetric imagery, and then applied a leaf index based segmentation to highlight the leaves region. In the end, the 3D model of each leaf can be represented and some geometric parameters of the leaf are designed and analyzed to predict the drought status. The experiments on real close-range stereo imagery justified the performance of the proposed approach to differentiate drought and healthy leaves.

Code and data availability

The paper describes a proprietary stereo image pair of a beech tree, a self-trained MC-CNN, and manual leaf selection, but contains no data availability statement, no public dataset deposit, and no code/model release. The only URLs are the paper's own DOI and cited external works (KITTI, Middlebury, MicMac), which are

No evidence-backed public reproduction asset is currently recorded.