Feature back-projection guided residual refinement for real-time stereo matching network

作者：

Highlights：

• We design a lightweight but efficient module to extract features. The module is composed of linear residual network, dilation convolution and spatial attention mechanism.

• We propose a lightweight 3D convolutional neural network with an hourglass structure to generate the initial disparity map.

• We propose a feature back-projection guided residual refinement module. This module uses a back-projection generator to generate high-frequency features to guide the disparity refinement.

• Experiments show that our proposed stereo matching network can achieve 25 fps on high-end GPU.

摘要

•We design a lightweight but efficient module to extract features. The module is composed of linear residual network, dilation convolution and spatial attention mechanism.•We propose a lightweight 3D convolutional neural network with an hourglass structure to generate the initial disparity map.•We propose a feature back-projection guided residual refinement module. This module uses a back-projection generator to generate high-frequency features to guide the disparity refinement.•Experiments show that our proposed stereo matching network can achieve 25 fps on high-end GPU.

论文关键词：Convolution neural networks,Feature back-projection,Real-time,Residual refinement,Stereo matching

论文评审过程：Received 18 February 2021, Revised 9 January 2022, Accepted 15 January 2022, Available online 20 January 2022, Version of Record 3 February 2022.

论文官网地址：https://doi.org/10.1016/j.image.2022.116636