15‐1: An image enhancement method for anomalous trichromats based on deep learning
SID Symposium Digest of Technical Papers · 2026-08 · 57 (S1) · 119-127
DOI: 10.1002/sdtp.70003_31 ↗Kurz gefasst
Fachartikel in der Zeitschrift SID Symposium Digest of Technical Papers, online erschienen am 11.09.2026. Verfasst von Zihao He und 3 weiteren Autor:innen. Die Arbeit nennt 34 Literaturangaben.
Abstract (Originalsprache)
In visual computing, making images easier to see for people with color vision deficiency (CVD), especially those with anomalous trichromacy, has always been a big challenge. Traditional color correction methods can make contrast stronger, but they often cause problems like colors blending into large patches or other color artifacts, so it's hard to keep things looking natural, particularly in severe cases. To address this, we propose a deep learning method based on Pix2Pix that integrates the Swin Transformer with a non‐local block guided by the red‐green difference. the shallow layers use local window attention to limit how much colors change, keeping things natural, while the deeper layers model global red‐green relationships to improve contrast. Compared to the traditional approaches from Woo et al. and Zhou et al., our method performs better on a test set with different severity levels, in both chromatic difference (CD) and Local Contrast Error (LCE) metrics. It effectively reduces colors blending into large patches and color distortion in severe CVD cases, while keeping more fine texture details intact. This work gives a practical way to improve digital media accessibility and viewing experience for people with color vision deficiencies.
Zitieren
APA
He, Z., Ma, Z., Zhang, B., & Ma, R. (2026). 15‐1: An image enhancement method for anomalous trichromats based on deep learning. SID Symposium Digest of Technical Papers, 57(S1), 119-127. https://doi.org/10.1002/sdtp.70003_31
Vancouver
He Z, Ma Z, Zhang B, Ma R. 15‐1: An image enhancement method for anomalous trichromats based on deep learning. SID Symposium Digest of Technical Papers. 2026;57(S1):119-127. doi:10.1002/sdtp.70003_31
BibTeX
@article{He2026,
author = {He, Zihao and Ma, Zejun and Zhang, Bingkun and Ma, Ruiqing},
title = {15‐1: An image enhancement method for anomalous trichromats based on deep learning},
journal = {SID Symposium Digest of Technical Papers},
year = {2026},
volume = {57},
number = {S1},
pages = {119--127},
doi = {10.1002/sdtp.70003_31}
}
EUVIMED