Depth map obtaining method and system, unmanned aerial vehicle orthogram generation method and medium

Abstract

The invention discloses a depth map obtaining method and system, an unmanned aerial vehicle orthogram generation method and a medium, and relates to the field of unmanned aerial vehicle remote sensingimage processing, and the method comprises the steps of photographing the same scene at a plurality of positions through an unmanned aerial vehicle, and obtaining a plurality of pictures; through feature point matching, obtaining a plurality of matching pairs among different pictures, and judging whether the pictures are overlapped or not; based on a matching result, performing motion recovery calculation to obtain camera external parameters and sparse space point clouds of all pictures; performing multi-view solid geometry calculation on all the pictures to obtain dense depth maps corresponding to all the pictures; according to the invention, block matching is used for matching between pixels, and spatial constraints are added to pixel point depth measurement through regularization processing of the block matching; according to the invention, the error of depth measurement can be reduced through space regularization constraint, the local depth tends to be uniform, the method is moresuitable for the situation that the depth change amplitude in a high-altitude unmanned aerial vehicle three-dimensional reconstruction scene is not large, and meanwhile the depth of some block matching failure points can be reasonably recovered.

Description

technical field [0001] The invention relates to the field of UAV remote sensing image processing, in particular, to a method and system for obtaining a depth map, a method and a medium for generating an UAV orthophoto map. Background technique [0002] At present, 3D reconstruction technology is becoming more and more mature, and the integration with deep learning is also in full swing. The most important technologies in 3D reconstruction are matching point pairs, SfM (motion recovery) and MVS (multi-view stereo vision). The main function of matching point pairs is to match pixels with the same name, and the mainstream methods are mainly feature point method and optical flow method. SfM estimates the camera parameters from the results of matching point pairs, and optimizes the camera parameters globally to obtain a smaller error globally. MVS uses the estimated camera parameters to measure the depth of the pixels in the image to obtain a dense or semi-dense point cloud. W...

AI Technical Summary

Problems solved by technology

In the traditional MVS method, the farther the distance, the less accurate the depth measurement

At the same time, it is necessary to adapt to the situation where the depth changes sharply and gently, but this increases the situation where measurement errors are prone to occur in the 3D reconstruction of the UAV

In addition, common block matching algorithms are also prone to mis-match

Images