Optimized highway deep learning network for fast single image super-resolution reconstruction

Ha, Viet Khanh and Ren, Jinchang and Xu, Xinying and Liao, Wenzhi and Zhao, Sophia and Ren, Jie and Yan, Gaowei (2020) Optimized highway deep learning network for fast single image super-resolution reconstruction. Journal of Real-Time Image Processing.

[img] Text (Khanh-Ha-etal-JRTIP-2020-Optimised-highway-deep-learning-network-for-fast)
Khanh_Ha_etal_JRTIP_2020_Optimised_highway_deep_learning_network_for_fast.pdf
Accepted Author Manuscript
Restricted to Repository staff only until 27 April 2021.

Download (5MB) | Request a copy from the Strathclyde author

    Abstract

    With the success of the deep residual network for image recognition tasks, the residual connection or skip connection has been widely used in deep learning models for various vision tasks, including single image super-resolution (SISR). Most existing SISR approaches pay particular attention to residual learning, while few studies investigate highway connection for SISR. Although skip connection can help to alleviate the vanishing gradient problem and enable fast training of the deep network, it still provides the coarse level of approximation in both forward and backward propagation paths and thus challenging to recover high-frequency details. To address this issue, we propose a novel model for SISR by using highway connection (HNSR), which composes of a nonlinear gating mechanism to further regulate the information. By using the global residual learning and replacing all local residual learning with designed gate unit in highway connection, HNSR has the capability of efficiently learning different hierarchical features and recovering much more details in image reconstruction. The experimental results have validated that HNSR can provide not only improved quality but also less prone to a few common problems during training. Besides, the more robust and efficient model is suitable for implementation in real-time and mobile systems.