Local Stereo Matching Using Adaptive Cross-Region-Based Guided Image Filtering with Orthogonal Weights

Adaptive cross-region-based guided image filtering (ACR-GIF) is a commonly used cost aggregation method. However, the weights of points in adaptive cross-region (ACR) are generally not considered, which affects accuracy disparity results. In this study, we propose an improved method to address issue. First, orthogonal weight proposed according structural feature ACR, and then each point ACR computed. Second, matching volume filtered using ACR-GIF with (ACR-GIF-OW). order reduce computing time method, efficient weighted based on proposed. Additionally, by combining ACR-GIF-OW our recently computation refinement local stereo algorithm as well. The results Middlebury evaluation platform show that, compared ACR-GIF, can significantly improve less additional overhead, performance outperforms other state-of-the-art nonlocal algorithms.