This is the authors' abstract. We don't add key points for this paper.
针对半全局立体匹配方法在匹配中容易受采样噪声的影响,以及在弱纹理区域匹配效果较差的问题,提出多尺度代价融合的立体匹配方法。首先构建图像金字塔,对每一层级进行代价计算,采用基于曼哈顿距离的SBT(Spatially-weighted Birchfield-Tomasi)代价、梯度代价、多灰度自适应阈值的Census代价的融合代价方法;然后对每一层级图像进行代价聚合,采用十字交叉区域代价聚合和扫描线优化代价聚合;接着对每一层级代价进行融合,得到代价融合的成本体;最后进行视差计算及优化。所提算法与传统半全局立体匹配算法AD-Census相比误匹配率减少15%。对算法进行CUDA的GPU加速,相比在CPU上运行速率提升超40倍,满足实时场景需求。
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Bin et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: