Spatially Adaptive De-Normalization Based Shadow Removal from a Single Image 


Vol. 47,  No. 4, pp. 584-593, Apr.  2022
10.7840/kics.2022.47.4.584


PDF
  Abstract

Shadow removal from a single-image has been a significant issue in image processing and many computer vision areas. This paper proposes a novel network based on the conditional generative adversarial network scheme without requiring additional shadow detection process to remove shadows. The proposed structure also utilizes a spatially adaptive de-normalization method to prevent the input image information loss caused by various normalization layers in the neural network. From the various experiments related to shadow removal using authorized datasets, it is confirmed that the proposed network shows at least 5㏈ higher performance in PSNR, compared to the state of the arts neural network based methodologies.

  Statistics
Cumulative Counts from November, 2022
Multiple requests among the same browser session are counted as one view. If you mouse over a chart, the values of data points will be shown.


  Cite this article

[IEEE Style]

H. Ryu and Y. Choe, "Spatially Adaptive De-Normalization Based Shadow Removal from a Single Image," The Journal of Korean Institute of Communications and Information Sciences, vol. 47, no. 4, pp. 584-593, 2022. DOI: 10.7840/kics.2022.47.4.584.

[ACM Style]

Hyunjeong Ryu and Yoonsik Choe. 2022. Spatially Adaptive De-Normalization Based Shadow Removal from a Single Image. The Journal of Korean Institute of Communications and Information Sciences, 47, 4, (2022), 584-593. DOI: 10.7840/kics.2022.47.4.584.

[KICS Style]

Hyunjeong Ryu and Yoonsik Choe, "Spatially Adaptive De-Normalization Based Shadow Removal from a Single Image," The Journal of Korean Institute of Communications and Information Sciences, vol. 47, no. 4, pp. 584-593, 4. 2022. (https://doi.org/10.7840/kics.2022.47.4.584)

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880