SuperMVS: Non-Uniform Cost Volume For High-Resolution Multi-View Stereo
This addresses the bottleneck of computational efficiency and accuracy in 3D reconstruction for computer vision applications, representing a novel method rather than an incremental improvement.
The paper tackles the problem of high computational cost and limited accuracy in multi-view stereo by proposing a non-uniform cost volume method with free-moving hypothesis planes, which reduces the number of planes while improving sampling fineness. It achieves state-of-the-art results with low memory, runtime, and fewer planes on DTU and Tanks & Temples datasets.
Different from most state-of-the-art~(SOTA) algorithms that use static and uniform sampling methods with a lot of hypothesis planes to get fine depth sampling. In this paper, we propose a free-moving hypothesis plane method for dynamic and non-uniform sampling in a wide depth range to build the cost volume, which not only greatly reduces the number of planes but also finers sampling, for both of reducing computational cost and improving accuracy, named Non-Uniform Cost Volume. We present the SuperMVS network to implement Multi-View Stereo with Non-Uniform Cost Volume. SuperMVS is a coarse-to-fine framework with four cascade stages. It can output higher resolution and accurate depth map. Our SuperMVS achieves the SOTA results with low memory, low runtime, and fewer planes on the DTU datasets and Tanks \& Temples dataset.