照片重建的主要步骤:
1.找出各张图片中的特征点,进行两两匹配;
要求能够精确识别物体的局部特征,并且进行快速准确的匹配。现常用的算法是由Dacid Lowe提出的SIFT方法。
2.根据匹配结果,利用射影定理计算得到相机位置等场景信息;
此步又称运动恢复结构(Structure from Motion),或稀疏重建(Sparse Reconstruction)。结果的衡量标准注意是准确性,现常用是基于Lecenberg-Marquardt算法的Bundler。
3.运用场景信息与原始照片,得到照片中物体的3D点云;
此步又称密集重建(Dense Reconstruction)。运用多视立体重建(Multi-view Stereo Reconstruction),得到3D点云。点云质量受到处理图像精度的执行效率、重建精度和完整性影响,目前最好的算法是PMVS。
4.根据3D点云构建3D模型;
将点连成面,才可在一般三维建模软件中使用。现常用的是泊松表面重建算法(Possion Surface Reconstruction)。
VisualSFM的基本理念是SFM(Structure From Motion):
Structure from motion (SfM) is a range imaging technique; it refers to the process of estimating three-dimensional structures from two-dimensional image sequences which may be coupled with local motion signals. It is studied in the fields of computer vision and visual perception. In biological vision, SfM refers to the phenomenon by which humans (and other living creatures) can recover 3D structure from the projected 2D (retinal) motion field of a moving ob