Abstract:With the development of remote sensing satellite technology for Earth observation, remote sensing stereo images have been used for three-dimensional reconstruction in various fields, such as urban planning and construction. However, remote sensing images often contain noise, occluded regions, untextured areas, and repeated textures, which can lead to reduced accuracy in stereo matching and affect the quality of 3D reconstruction results. To reduce the impact of complex scenes in remote sensing images on stereo matching and to ensure both speed and accuracy, we propose a new end-to-end stereo matching network based on convolutional neural networks (CNNs). The proposed stereo matching network can learn features at different scales from the original images and construct cost volumes with varying scales to obtain richer scale information. Additionally, when constructing the cost volume, we introduce negative disparity to adapt to the common occurrence of both negative and non-negative disparities in remote sensing stereo image pairs. For cost aggregation, we employ a 3D convolution-based encoder–decoder structure that allows the network to adaptively aggregate information. Before feature aggregation, we also introduce an attention module to retain more valuable feature information, enhance feature representation, and obtain a higher-quality disparity map. By training on the publicly available US3D dataset, we obtain an accuracy of 1.115 pixels in end-point error (EPE) and 5.32% in the error pixel ratio (D1) on the test dataset, and the inference speed is 92 ms. Comparing our model with existing state-of-the-art models, we achieve higher accuracy, and the network is beneficial for the three-dimensional reconstruction of remote sensing images.

Parallax Attention for Unsupervised Stereo Correspondence Learning

Multi-scale Parallax Attention for Stereo Image Super-Resolution

Parallax attention stereo matching network based on the improved group-wise correlation stereo network

Stereo Image Restoration Via Attention-Guided Correspondence Learning

Coarse-to-Fine Cross-View Interaction Based Accurate Stereo Image Super-Resolution Network

Learning Inter- and Intra-frame Representations for Non-Lambertian Photometric Stereo

Superpixel Guided Network for Three-Dimensional Stereo Matching

Multi-Dimensional Attention on Cost Volume for Stereo Matching

Practical Stereo Matching via Cascaded Recurrent Network with Adaptive Correlation

Capturing Cross-Scale Disparity for Stereo Image Super-Resolution

Stereo Matching Method for Remote Sensing Images Based on Attention and Scale Fusion

Multi-Scale Context Attention Network for Stereo Matching

A Joint 2D-3D Complementary Network for Stereo Matching

Stereo Matching by Self-supervision of Multiscopic Vision.

A Fast Stereo Matching Network with Multi-Cross Attention

Adaptive Cost Volume Representation for Unsupervised High-resolution Stereo Matching

Deep Stereo Matching With Hysteresis Attention and Supervised Cost Volume Construction

End-to-End Learning of Multi-scale Convolutional Neural Network for Stereo Matching

Self-Supervised Stereo Matching Method Based on SRWP and PCAM for Urban Satellite Images

Stereo Matching Using Multi-Level Cost Volume and Multi-Scale Feature Constancy

Robust Depth Estimation Based on Parallax Attention for Aerial Scene Perception