基于自适应空间稀疏化的高效多视图立体匹配

To reduce the high computational complexity in constructing and aggregating cost volumes for multi-view stereo matching, existing methods commonly employ cascaded architectures or iterative optimization. However, these ap⁃ proaches still face two main challenges. The cascaded architectures narrow down the depth sampling range during the re⁃ finement stage, which may lead to erroneous estimation of depth discontinuities. While the inference time of iterative opti⁃ mization networks linearly increases with the number of iterations, making it difficult to meet the requirements of real-time systems. To address these challenges, this paper proposes an efficient multi-view stereo matching network via adaptive spa⁃ tial sparsification. We introduce a sparse matching cost volume that sparsely samples within the complete depth range, re⁃ ducing computational complexity while maintaining the network's ability to model depth-discontinuous regions. Mean⁃ while, we propose a sparse iterative optimization method that progressively prunes regions with converged depth values dur⁃ ing iterations using adaptive variational Dropout, resulting in sub-linear growth in inference time with iteration count. Ex⁃ perimental results on the public datasets, DTU and Tanks & Temples, demonstrate that the proposed method achieves 1.2× and 0.35× improvements of inference speed compared to CasMVSNet and PatchmatchNet, respectively. Moreover, it exhib⁃ its excellent performance in point cloud reconstruction, effectively handles details in depth-discontinuous regions, and dem⁃ onstrates outstanding generalization capability.