Journal of Engineering, Project, and Production Management, 2026, 16(6), 2026-0077
Color Transfer for Map Faceted Elements Based on an Enhanced Generative Adversarial Network with Attention Mechanisms
1 Associate
Professor, The School of Visual Art, Hunan Mass Media Vocational and
Technical College, Changsha, 410000, China, E-mail: dzengd@outlook.com
(corresponding author).
2 Professor, The School of Visual
Art, Hunan Mass Media Vocational and Technical College, Changsha,
410000, China
Project Management
Received June 2, 2026; revised July 31, 2026; accepted August 8, 2026
Available online August 30, 2026
Abstract: Color transfer for map faceted elements in cartography has long relied on manual efforts, which are inefficient and highly subjective, making it difficult to balance color accuracy with structural fidelity. This study integrates generative adversarial networks with two types of attention mechanisms to propose a color transfer scheme tailored for map aerial elements. The method optimizes the generator using residual units to preserve spatial details and employs gradient normalization to enhance the discriminator's training stability. Channel attention governs the overall tonal distribution, whereas spatial attention focuses on regional boundaries, ensuring that the transferred results align closely with real maps in both color and shape. To evaluate performance, we used three metrics: mean Intersection over Union (mIoU), pixel error, and visual quality metrics. The mIoU measures semantic consistency in color distribution, while pixel error quantifies color reproduction accuracy. Peak Signal‑to‑Noise Ratio (PSNR) and Structural Similarity Index (SSIM) assess visual quality. Experimental results on the Urban Green Space (UGS) dataset show strong performance. The proposed method achieves an mIoU of 0.81, a pixel error of 4.84%, and a processing time of 0.09 seconds per image. It also achieves a PSNR of 34.17 dB and an SSIM of 0.91. Comparable performance is observed on the EnvMonitor dataset. This technique effectively maintains semantic consistency in color distribution, spatial boundary precision, and cartographic standards while ensuring high transfer efficiency, offering a novel pathway for automated map color design that reconciles speed and accuracy.
Keywords: Improved generative adversarial networks; attention mechanisms; map faceted color transfer; residual networks; gradient normalization.
Copyright © Journal of Engineering, Project, and Production Management (EPPM-Journal).
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.
Requests for reprints and permissions at eppm.journal@gmail.com.
Citation: Xie, B. and Zeng, D. (2026). Color Transfer for Map Faceted Elements Based on an Enhanced Generative Adversarial Network with Attention Mechanisms. Journal of Engineering, Project, and Production Management, 16(6), 2026-0077.
