Tree modeling method and tree modeling device based on single image and equipment

Abstract

The invention discloses a tree modeling method and tree modeling device based on a single image and equipment. The tree modeling method comprises the following steps: acquiring a tree edge graph and a visible branch graph of a single tree image to be reconstructed; inputting the tree edge map and the visible branch map into a pre-trained conditional generative adversarial network to obtain a contour depth map and a branch depth map of the tree; obtaining contour point cloud data according to the contour depth map, and constructing a branch three-dimensional skeleton according to the branch depth map; and by taking the contour point cloud data and the branch three-dimensional skeleton as constraint conditions, generating a three-dimensional tree model of the to-be-reconstructed single tree image according to a predetermined self-organizing process type modeling method. A good modeling effect can be achieved for trees with irregular contours, depth data of the tree contours can be accurately obtained through the conditional generative adversarial network, and information of missing tree detail branches can be recovered by adopting the process type modeling method.

Description

technical field [0001] The invention belongs to the technical field of computer graphics, and in particular relates to a tree modeling method based on a single image, a tree modeling device, a computer-readable storage medium, and computer equipment. Background technique [0002] Many tree modeling methods have been proposed in the past few years, including procedural tree modeling and reconstruction of tree models from photographs or laser scans. Procedural tree modeling synthesizes virtual tree models, while tree reconstruction methods generate real-world tree models. To reconstruct a realistic model of a real tree, it is costly and time-consuming to collect input data, such as using a laser scanner to obtain a 3D point cloud or carefully taking numerous photos from different views around the real tree. Single-image-based tree reconstruction techniques take only one image as input, which greatly simplifies the input data requirements for real tree reconstruction, thus pro...

AI Technical Summary

Problems solved by technology

This image-based modeling method uses very simple 2D to 3D transformation rules, specifically it can only model very regular plants well due to the assumption of similarity in front and side profiles effect, good results cannot be achieved for plants with very different front and side profiles

Images