基于多粒度匹配的高超速飞行体三维重构方法

3D Reconstruction Methods of Hypersonic Object Based on Multi-granularity Matching

摘要:

针对高超速飞行体在飞行过程中能获取到的图像信息较少,无法完全复现其轮廓信息,提出一种基于多粒度匹配的三维重构优化方法。首先通过两台高分辨率CCD相机正交的阴影照相站系统采集高超速飞行体多幅图像信息,利用图像分割技术提取物体的二维轮廓；然后采用改进的SFS算法求解出物体表面各点的相对高度和表面法向量,恢复其三维形貌；最后使用图像拼接优化技术实现高超速飞行体表面的三维重构。并经过实体模型的三维重构验证所提方法的可行性及有效性。

Abstract:

As an important problem in 3D reconstruction of hypersonic object, it is difficult to acquire images during the flight, and its contour information cannot be completely rebuilt.3D reconstruction methods of hypersonic object based on multi-granularity matching is adopted. First, multiple images of hypersonic object can be collected by orthogonal shadowgraph system, which is composed of two high resolution Charge Coupled Device (CCD) cameras, and the 2D profile will be extracted using image segmentation technique. Then, surface normal vector and relative heights of each surface point are solved by improved Shape Form Shading (SFS) method, and the 3D shape information will be restored. At last, 3D reconstruction of hypersonic object surface is realized by optimized image stitching. And feasibility and validity of the method is verified by 3D reconstruction experiment.