Tree three-dimensional reconstruction method based on unmanned aerial vehicle aerial photo sequence image

Abstract

The invention discloses a tree three-dimensional reconstruction method based on unmanned aerial vehicle aerial photo sequence images. According to the tree three-dimensional reconstruction method, firstly, characteristic regions of a crown are extracted in a watershed partition method, then, the characteristic regions are matched by calculating region correlation coefficients in an RGB color space, matched characteristic dot pairs are extracted, depth information is calculated through the binocular stereo vision principle combined with imaging models of an aerial photo camera, and finally a three-dimensional model of a tree is constructed based on L system laws. Compared with the prior art, the tree three-dimensional reconstruction method has the advantages that grain, color and outline information of the crown are considered, and the characteristic spot set which can show the crown complicated structure is fully extracted; in addition, the depth information of the characteristic spot set is calculated according to complicated imaging models of the aerial photo camera with shake, transverse moving and the like of the camera in the aerial photo process of an unmanned aerial vehicle considered, a reasonable tree modeling approach is adopted, and therefore reasonable approximate three-dimensional models of the tree are constructed.

Description

technical field [0001] This patent relates to image three-dimensional reconstruction technology, in particular to a three-dimensional reconstruction method based on UAV aerial photography sequence images. Background technique [0002] UAV aerial photography is an important data source for spatial data acquisition in recent years. UAV aerial photography can perform multiple imaging of the same target and scene to form multi-view and multi-position imaging. Through the onboard high-resolution optical camera, the UAV can obtain a large number of image sequences at one time, and these images have high resolution, high overlap rate, and can be measured. Applying the relevant theories and technologies of 3D reconstruction of sequence images, the 3D model of the ground target can be obtained by processing the UAV sequence imaging. [0003] The principle of the three-dimensional image reconstruction method is to obtain two images of the same target from different angles through th...

AI Technical Summary

Problems solved by technology

In the 3D reconstruction of aerial images, the vibration, rotation, and movement of the drone's aerial photography will cause the vibration of the aerial camera, which makes the imaging model of the aerial camera more complicated.

Among them, most of the 3D reconstruction of tree objects using the existing image 3D reconstruction methods only model the backbone structure information of the tree, and rarely consider the tree crown outline and internal texture details, etc., which often lead to the reconstructed 3D image distortion of tree objects

Images