Local Similarity Pattern and Cost Self-Reassembling for Deep Stereo Matching Networks

TL;DR

This paper introduces a Local Similarity Pattern and a dynamic self-reassembling refinement strategy to enhance deep stereo matching networks, addressing limitations in feature discrimination and over-smoothing, with demonstrated improvements on benchmark datasets.

Contribution

The paper proposes novel LSP features and a self-reassembling refinement method, integrating traditional ideas into deep stereo matching to improve accuracy and reduce over-smoothing.

Findings

Significant performance improvements on SceneFlow and KITTI benchmarks.

Enhanced feature discrimination through LSP for better matching accuracy.

Reduced over-smoothing in disparity maps with the proposed refinement strategy.

Abstract

Although convolution neural network based stereo matching architectures have made impressive achievements, there are still some limitations: 1) Convolutional Feature (CF) tends to capture appearance information, which is inadequate for accurate matching. 2) Due to the static filters, current convolution based disparity refinement modules often produce over-smooth results. In this paper, we present two schemes to address these issues, where some traditional wisdoms are integrated. Firstly, we introduce a pairwise feature for deep stereo matching networks, named LSP (Local Similarity Pattern). Through explicitly revealing the neighbor relationships, LSP contains rich structural information, which can be leveraged to aid CF for more discriminative feature description. Secondly, we design a dynamic self-reassembling refinement strategy and apply it to the cost distribution and the disparity map respectively. The former could be equipped with the unimodal distribution constraint to alleviate the over-smoothing problem, and the latter is more practical. The effectiveness of the proposed methods is demonstrated via incorporating them into two well-known basic architectures, GwcNet and GANet-deep. Experimental results on the SceneFlow and KITTI benchmarks show that our modules significantly improve the performance of the model.