As one of the well-known algorithms in the field of image restoration, Multi-Stage Progressive Image Restoration (MPR-Net) achieved quite great results in the fields of computer vision and image denoising. MPR-Net introduces the attention mechanism into the traditional encoder-decoder network and integrates the idea of multi-stage learning, so that the network can better learn the feature information of the image and improve the effect of image restoration. This paper takes the MPR-Net model as the main research object to introduce the principle of MPR-Net. In addition, this paper adopts a self-made binocular camera to collect the data set, and the degraded images in the data set are actually captured, which is different from the public GOPRO data set, which uses an average of 15 frames of clear images to obtain degraded images. The MPR-Net model is successfully applied to the self-made data set to realize image restoration, which verifies the effectiveness and portability of the MPR-Net model. Meanwhile, compared with the GOPRO data set, the image features of the self-made data set in this paper are seriously degraded, and the restoration effect of the MPR-Net model on this data set is greatly reduced. In this paper, the peak signal-to-noise ratio (PSNR) and structural similarity (SSIM) are used as measurement indicators. The training result of the MPR-Net model on the self-made dataset is 0.7304 and the PSNR is 25.79dB, which is 0.2 and 5.96dB less than the training result on the public GOPRO dataset, respectively.

Jingyao Zhang. Research on portability and effectiveness based on MPR-Net. Academic Journal of Computing & Information Science (2024), Vol. 7, Issue 11: 27-33. https://doi.org/10.25236/AJCIS.2024.071104.

[1] Syed Waqas Zamir, Aditya Arora, Salman Khan, Munawar Hayat, Fahad Shahbaz Khan, Ming-Hsuan Yang, and Ling Shao. Multi-Stage Progressive Image Restoration [C]. arXiv:2102.02808, 2021. 

[2] E. Rublee, V. Rabaud, K. Konolige, and G. Bradski, ORB: an efficient alternative to SIFT or SURF [C]. In ICCV 2011.

[3] Chen Chao, Qi Feng. Review on the development of Convolutional neural networks and their applications in computer vision [J]. Computer science, 2019, 46 (3): 11. DOI: 10.11896 / j.i SSN. 1002- 137 - x. 2019.03.008.

[4] Saeed Anwar and Nick Barnes. Real image denoising with feature attention[C]. ICCV, 2019.

[5] Olaf Ronneberger, Philipp Fischer, Thomas Brox. U-Net: convolutional networks for biomedical image segmentation[C]. In MICCAI, 2015.

[6] Yulun Zhang, Kunpeng Li, Kai Li, Bineng Zhong, and Yun Fu. Residual non-local attention networks for image restoration[C]. In ICLR, 2019.

[7] Kong J Y, Zhang H S, Improved U-Net network and its application of road extraction in remote sensing image [J]. Chinese Science and Technology, 2022, 42(3): 105-113. (in Chinese)

[8] Chen Ruilin, Zhang Bo, Duan Xikai, et al. Remote Sensing Image Resolution Enhancement Technology Based on Single-Pixel Imaging [J]. Spacecraft Recovery & Remote Sensing, 2023, 44(6): 130-139. (in Chinese)