The knowledge of tilt photogrammetry that novices should know

The so-called "standing on the shoulders of giants" is well reflected in this article. Some of the pictures and concepts in this paper are based on the pictures and concepts released by "Giant". Thanks to these "giants" for leading us on a bright road. I'm not a professional in photogrammetry, and I'm just a novice in GIS circle. Please allow me to talk about the beginner's knowledge of tilt photogrammetry from the perspective of a novice.

In recent years, tilt photogrammetry has set off a huge wave in GIS. Some people say that it subverts the traditional surveying and mapping field, while others say that it replaces the traditional modeling method. It is precisely because tilt photogrammetry has been brought with so many auras that people begin to have various conjectures about it, making it more and more mysterious. Let's take a look at the renderings of oblique photography data.

Let's take a look at tilt photogrammetry first. What is the difference between it and our traditional image?

From the way of data collection, the traditional image is the image data that is continuously photographed on the ground by the aerial camera carried on the plane and then processed in a series of rooms, and the only result is the information of the overlooking angle of the ground object, that is, the viewing angle is vertical to the ground. In the tilt photogrammetry experiment, an airplane or unmanned aerial vehicle is equipped with five cameras to shoot ground objects from four vertical directions, and then through a series of indoor processing such as geometric correction, adjustment and multi-view image matching, data with all-round information of ground objects are obtained. The simple understanding is that the ground objects in the image are on a plane, and the ground objects measured by oblique photography have real heights.

We know the data collection method of tilt photography. Through the joint adjustment of multi-view images, key matching of multi-view images, digital curved surface model making, real projection image correction and other key technologies of oblique photography data processing, more edge information of curved surface data can be obtained, and the three-dimensional model of the data can be obtained through indoor data processing.

Next, let's unveil the mystery of tilt photogrammetry data! The data of tilt photogrammetry is essentially a grid model. What is a grid model? Mesh model is a kind of mesh surface model, which is composed of point clouds through some algorithms, such as region growing method, octree algorithm, wavefront algorithm and so on. Point cloud is a massive collection of points used to represent the spatial distribution and surface characteristics of targets in the same spatial reference system. Based on geometric correction and aberration processing flow, office software can calculate ultra-high density point clouds from images, as shown in the following figure:

The above two pictures show that the cutting effect is quite good and orderly, but when enlarged to a certain extent, the effect is as follows: