DOI QR코드

DOI QR Code

질감 대조 가중치를 이용한 단일 영상의 초해상도 기법

Single Image Super Resolution Method based on Texture Contrast Weighting

  • Hyun Ho Han (College of General Education, University of Ulsan)
  • 투고 : 2024.02.10
  • 심사 : 2024.03.20
  • 발행 : 2024.03.28

초록

본 논문은 초해상도 결과의 품질을 향상시키기 위해 질감 특징을 세분화하여 각각을 대조하고, 그 결과를 가중치로 이용하는 초해상도 방법을 제안하였다. 초해상도에서 중요한 평가 기준인 품질의 향상을 위해서는 경계 영역과 같은 세부사항에서의 정확하고 명확한 복원 결과가 필요하며, 인공물과 같은 불필요한 잡음을 최소화하는 것이 중요하다. 제안하는 방법은 품질 향상을 위해 기존 CNN(Convolutional Neural Network) 기반의 초해상도 방법에서 특징 추정을 위해 다중 경로의 잔차 블록 구조와 skip-connection을 구성하였다. 추가적인 질감 분석을 위한 선명 및 흐림 이미지 결과를 추가로 학습하였다. 이를 활용하여 초해상도 수행 결과 또한 각각을 대조하여 가중치를 할당하는 방법을 이용해 영상의 세부사항 영역과 평활화 영역에 대해 개선된 품질을 얻을 수 있도록 하였다. 제안하는 방법의 실험 결과 평가 기준으로 활용되는 PSNR과 SSIM 값이 기존 알고리즘 대비 높은 결과 값을 얻어 품질이 개선됨을 확인할 수 있었다.

In this paper, proposes a super resolution method that enhances the quality of results by refining texture features, contrasting each, and utilizing the results as weights. For the improvement of quality, a precise and clear restoration result in details such as boundary areas is crucial in super resolution, along with minimizing unnecessary artifacts like noise. The proposed method constructs a residual block structure with multiple paths and skip-connections for feature estimation in conventional Convolutional Neural Network (CNN)-based super resolution methods to enhance quality. Additional learning is performed for sharpened and blurred image results for further texture analysis. By contrasting each super resolution result and allocating weights through this process, the proposed method achieves improved quality in detailed and smoothed areas of the image. The experimental results of the proposed method, evaluated using the PSNR and SSIM values as quality metrics, show higher results compared to existing algorithms, confirming the enhancement in quality.

키워드

참고문헌

  1. K. Li, S. Yang, R. Dong, X. Wang & J Huang. (2020). Survey of single image super-resolution reconstruction. IET Image Processing, 14(11), 2273-2290. DOI : 10.1049/iet-ipr.2019.1438 
  2. Z. Lu, J. Li, H. Liu, C. Huang, L. Zhang & T. Zeng (2022). Transformer for single image super-resolution. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 457-466). DOI : 10.48550/arXiv.2108.11084 
  3. K. Singla, R. Pandey & U. Ghanekar. (2022). A review on Single Image Super Resolution techniques using generative adversarial network. Optik, 266, 169607. DOI : 10.1049/iet-ipr.2019.1438 
  4. G. Wu, J. Jiang & X. Liu. (2023). A practical contrastive learning framework for single-image super-resolution. IEEE Transactions on Neural Networks and Learning Systems. DOI : 10.1109/TNNLS.2023.3290038 
  5. K Chauhan et al. (2023). Deep learning-based single-image super-resolution: a comprehensive review. IEEE Access, 11, 21811-21830. DOI : 10.1109/ACCESS.2023.3251396 
  6. C. Tian, Y. Yuan, S. Zhang, C. W. Lin, W. Zuo, D. Zhang. (2022). Image super-resolution with an enhanced group convolutional neural network. Neural Networks, 153, 373-385. DOI : 10.1016/j.neunet.2022.06.009 
  7. X. Zhang, H. Zeng, S. Guo & L. Zhang. (2022, October). Efficient long-range attention network for image super-resolution. In European conference on computer vision (pp. 649-667). Cham: Springer Nature Switzerland. DOI : 10.1007/978-3-031-19790-1_39 
  8. B. Liu & D. Ait-Boudaoud. (2020). Effective image super resolution via hierarchical convolutional neural network. Neurocomputing, 374, 109-116. DOI : 10.1016/j.neucom.2019.09.035 
  9. J. Kim, J. K. Lee & K. M. Lee. (2016). Accurate image super-resolution using very deep convolutional networks. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 1646-1654). DOI : 10.1109/CVPR.2016.182 
  10. J. Kim, J. K. Lee & K. M. Lee. (2016). Deeply-recursive convolutional network for image super-resolution. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 1637-1645). DOI : 10.1109/CVPR.2016.181 
  11. N. Ahn, B. Kang & K. A. Sohn. (2018). Fast, accurate, and lightweight super-resolution with cascading residual network. In Proceedings of the European conference on computer vision (ECCV) (pp. 252-268). DOI : 10.48550/arXiv.1803.08664 
  12. B. Lim, S. Son, H. Kim, S. Nah & K. Mu Lee (2017). Enhanced deep residual networks for single image super-resolution. In Proceedings of the IEEE conference on computer vision and pattern recognition workshops (pp. 136-144). DOI : 10.48550/arXiv.1707.02921 
  13. Y. Zhang, Y. Tian, Y. Kong, B. Zhong & Y. Fu (2018). Residual dense network for image super-resolution. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 2472-2481). DOI : 10.48550/arXiv.1802.08797 
  14. Z. Lu & Y. Chen. (2022). Single image super-resolution based on a modified U-net with mixed gradient loss. signal, image and video processing, 16(5), 1143-1151. DOI : 10.1007/s11760-021-02063-5 
  15. M. Zhang & Q. Ling. (2020). Supervised pixel-wise GAN for face super-resolution. IEEE Transactions on Multimedia, 23, 1938-1950. DOI : 10.1109/TMM.2020.3006414