Image super-resolution reconstruction based on sparse representation and deep learning

• Published in: Signal processing. Image communication Vol. 87; p. 115925
• Main Authors: Zhang, Jing, Shao, Minhao, Yu, Lulu, Li, Yunsong
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 01.09.2020; Elsevier BV
• Subjects: Algorithms; Deep learning; Feature extraction; Feature fusion; Gabor transformation; Image processing; Image quality; Image reconstruction; Image resolution; Image restoration; Sparse representation; Spatial resolution; Super-resolution; Deep learning; Sparse representation; Super-resolution; Feature fusion
• ISSN: 0923-5965; 1879-2677
• DOI: 10.1016/j.image.2020.115925

Super-resolution reconstruction technology has important scientific significance and application value in the field of image processing by performing image restoration processing on one or more low-resolution images to improve image spatial resolution. Based on the SCSR algorithm and VDSR network, in order to further improve the image reconstruction quality, an image super-resolution reconstruction algorithm combined with multi-residual network and multi-feature SCSR(MRMFSCSR) is proposed. Firstly, at the sparse reconstruction stage, according to the characteristics of image blocks, our algorithm extracts the contour features of non-flat blocks by NSCT transform, extracts the texture features of flat blocks by Gabor transform, then obtains the reconstructed high-resolution (HR) images by using sparse models. Secondly, according to improve the VDSR deep network and introduce the feature fusion idea, the multi-residual network structure (MR) is designed. The reconstructed HR image obtained by the sparse reconstruction stage is used as the input of the MR network structure to optimize the high-frequency detail residual information. Finally, we can obtain a higher quality super-resolution image compared with the SCSR algorithm and the VDSR algorithm. •The innovation of this thesis lies in the combination of sparse coding and deep learning.•We use the Gabor transform and NSCT transform to propose the MFSCSR algorithm, which is better than the SCSR algorithm.•We improve the VDSR deep network and introduce the feature fusion idea, and propose the multi-residual network.•We combine the MFSCSR algorithm with the MR network to obtain the MRMFSCSR algorithm, which has better image reconstruction effect than the VDSR algorithm.