TL;DR本文提出了一种新的神经重建方法,使用可访问的 2D 图像作为监督来进行 3D 场景重建,其中通过引入二进制带注符号距离函数、色彩场和概率场等方法实现了场景的重建,并引入了几何和平面约束以实现对场景复杂几何区域和低纹理区域的重建。
Abstract
neural scene reconstruction methods have achieved impressive performance in reconstructing complex geometry and low-textured regions in large scenes. However, these methods heavily rely on 3D supervised informati