Image matching and 3D scene generation method based on UAV remote sensing

Aiming at unmanned aerial vehicles (UAV) remote sensing image matching and 3D model generation for complex scenes, an improved speeded-up robust features (SURF) feature matching algorithm is firstly proposed, which introduces sub-pixel level positioning and feature point orientation correction techniques, significantly improving the accuracy and robustness of feature point matching. On this basis, the principles of UAV remote sensing photogrammetry are summarized, the obtaining method of remote sensing images from UAVs and the key technologies for 3D scene modeling using software are also explored based on the ContextCapture real scene modeling software. Finally, typical scene generation experiments were performed on a self-built dataset. The experimental results show that the proposed image matching method in this paper has a significantly improvement in accuracy comparing to the traditional SURF method, and it has strong adaptability to changes in the scale, brightness and viewing angle of the image; the experimental results of the 3D reconstruction of the realistic remote sensing image based on UAV also show that the model has a high accuracy, and performs well in terms of the texture resolution, texture mapping accuracy, texture details and quality of 3D scene reconstruction.