A Matching Method for Oblique Photography 3D Reconstruction

Abstract

The invention relates to the field of three-dimensional reconstruction of unmanned aerial vehicle oblique photography, and discloses a three-dimensional reconstruction matching method of oblique photography, including extracting image position and pose feature components; extracting image color feature components; extracting texture feature components; and standardizing the three components Combined after processing to obtain the fusion feature vector; construct a hash function to map the obtained fusion feature vector; further calculate the Euclidean distance and sort to form a set of images to be matched, and then the single image matching object in the process of motion restoration structure from the to-be-matched Select from the image collection. Compared with the prior art, the present invention integrates UAV image POS information and color texture information, and clusters the pre-reconstructed image according to the characteristics of the UAV image, avoiding the global matching process of the traditional reconstruction algorithm. The reconstruction efficiency is significantly improved, and while the reconstruction efficiency is improved, the reconstructed dense point cloud of the present invention can maintain good accuracy.

Description

technical field [0001] The invention relates to the technical field of three-dimensional reconstruction of oblique photography of unmanned aerial vehicles, in particular to a method for three-dimensional reconstruction and matching of oblique photography. Background technique [0002] In recent years, with the continuous progress and perfection of feature point detection and matching algorithms, self-calibration algorithms, structure from motion reconstruction algorithms (Structure from motion, SFM) and multi-view stereo matching algorithms (Multi-view stereopsis, MVS), Image-based 3D reconstruction technology has developed by leaps and bounds, and has been widely used in forest resource survey, forest stand research and 3D reconstruction of standing trees. Due to the low-altitude flight characteristics of UAVs, it has the advantages of flexible viewing angles and strong timeliness. UAVs can be used to obtain a large number of uncalibrated image sequences of continuous viewi...

AI Technical Summary

Problems solved by technology

Xu Zhihua et al. proposed a method for constructing image topological relationship based on GPS information, which limited the image matching in the 3D reconstruction process to images with topological relationship, which reduced the time-consuming matching, but did not take into account that the camera position is the same, shooting completely different angle

Song Zhengxi et al. divided the image into blocks and constructed an image KD-tree by extracting Sift features before matching. However, the initial image clustering still uses Sift features, and the clustering efficiency is not very high.

Images