基于逐层优化SGM的倾斜影像三维重建

3D reconstruction of oblique image based on layer-by-layer optimization of semi-global matching

基于传统的半全局匹配(SGM)进行无人机倾斜影像三维重建时,匹配效率较低且容易产生误匹配,而逐层优化SGM是提高倾斜影像三维重建精度和效率的有效方法。文章在无人机倾斜影像三维重建过程中采用逐层优化SGM,利用低分辨率金字塔的匹配结果限制高分辨率金字塔的视差搜索范围,对SGM的动态规划路径进行像素约束,并对倾斜影像像对及像组进行匹配和三维重建实验。结果表明:采用逐层优化SGM时,倾斜影像像对的匹配时间缩短了30%,且其结果中包括了更多的视差信息及更少的错误匹配点;以逐层优化SGM的匹配结果进行三维重建,提高了模型的精细化程度;在建筑物和相对平坦的道路、草地等区域拥有了更好的三维重建效果,而在林木及测区边缘区域的三维重建效果有所降低。

It is easy to generate mismatches when 3D reconstruction based on UAV oblique image is carried out based on traditional semi-global matching(SGM),and the matching efficiency is low.SGM based on layer-by-layer optimization is an effective method to improve the accuracy and efficiency of 3D reconstruction based on oblique image.In this article,SGM based on layer-by-layer optimization is used in the process of 3D reconstruction based on UAV oblique images,the matching result of the low resolution pyramid is used to limit the search range of high resolution pyramid,pixel constraints are applied to the dynamic planning paths of SGM,and the experiment of matching and 3D reconstruction based on oblique image pairs and oblique image groups is carried out.The results show that:by using SGM based on layer-by-layer optimization,the matching time based on oblique image pairs is reduced by 30%,and the matching results include more parallax information and fewer mismatching points.The 3D reconstruction based on the matching result of SGM based on layer-by-layer optimization improves the refinement of the model.It has better 3D reconstruction effect in buildings,relatively flat roads and grasslands.However,the effect reduces in the forest and the edge of the survey area.