Abstract:Current deep-learning-based stereo matching algorithms achieve remarkably low error rates but they suffer from the edge ambiguity effect. The primary reason is that they treat disparity estimation as a labeling problem, constructing a cost volume based on uniform discrete pixel-wise labels. It is insufficient to model the continuous disparity probability distribution (DPD), which harms the accuracy of complex regions. Moreover, current cost aggregation strategies cannot process unstructured disparity candidates very well, which is one of the bottlenecks limiting continuous modeling. We propose Patchmatch Stereo++, inspired by the traditional Patchmatch Stereo to achieve better continuous disparity optimization in deep-learning-based methods. Firstly, to model accurate continuous DPD, we introduce an adaptive dense sub-pixel sampling strategy to binocular stereo and approximate a continuous unstructured DPD for every pixel. Secondly, we design a convolution-based optimizer that can accept unstructured disparity candidates to parse the above continuous DPD in an adaptive manner and perform updates accordingly. Extensive experiments demonstrate our method has the best performance among existing stereo matching networks at the edges, both quantitatively and qualitatively. At the time of submission, compared with published works pre-trained on SceneFlow, we rank 1st in the foreground of KITTI and 2nd on SceneFlow, ETH3D under various metrics.The source code will be released.

A Progressive Framework for Dense Stereo Matching

Segment-based Continuous Stereo Matching

Space-temporal stereo matching for dynamic scene

Accurate Real-Time Stereo Correspondence Using Intra- and Inter-Scanline Optimization

Robust stereo matching with surface normal prediction.

Sparse LIDAR Measurement Fusion with Joint Updating Cost for Fast Stereo Matching

Temporally Consistent Stereo Matching

Dense Stereo Matching from Separated Views of Wide-Baseline Images

Stereo matching method based on high-precision occlusion-recovering and discontinuity-preserving

Segmentation-based dense stereo matching algorithm for virtual view synthesis

Dense Stereo Correspondence with Contrast Context Histogram, Segmentation-Based Two-Pass Aggregation and Occlusion Handling

Stereo Matching with Cost Volume based Sparse Disparity Propagation

Feature Based Stereo Matching Using Two-Step Expansion

A Stereo Matching Algorithm Based on Census Transform and Improved Dynamic Programming

Segment-Based Disparity Refinement With Occlusion Handling for Stereo Matching

A Segment-Based Dense Stereo Matching Algorithm

Adaptive Support-Weight Stereo-Matching Approach with Two Disparity Refinement Steps

A stereo matching algorithm based on selective belief propagation

Dense Stereo Matching Algorithm Based on Image Segmentation

Patchmatch Stereo++: Patchmatch Binocular Stereo with Continuous Disparity Optimization

Accurate edge-preserving stereo matching by enhancing anisotropy.