Scholarlink

<p>Graphic design is a visual communication tool that communicates messages using elements like text, imagery, color, and layout. Among these elements, accurate color element extraction is essential to maintain consistency, optimize color contrasts, and create visually engaging designs. Therefore, this paper introduces a Deep Graph Convolutional based Random Skill Algorithm (DGC-RSA) that combines Deep Graph Convolutional Networks (DGCN) and RSA to optimize color feature analysis in graphic design. Initially, pixel-based color extraction is performed for color clustering and the Median-Cut algorithm is used to perform color quantification. Secondly, perceptual characteristics such as hue, saturation, and brightness are analyzed globally and locally to optimize emotional and communicative impact. The core innovation lies in the DGCN which models color clusters as graph nodes and spatial relationships as edges to derive meaningful color patterns, thereby enabling the framework to capture intricate contextual dependencies and enhance design harmony. Moreover, the hyperparameters of the DGCN are fine-tuned using an RSA algorithm that enables efficient exploration and exploitation of the search space to ensure convergence by adapting dynamically and avoiding local optima through randomness. The comprehensive evaluation is conducted by using an Image colorization dataset using diverse metrics including, color contrast ratio, color accuracy, inception score, error rate, PSNR, and scalability. The results revealed that the DGC-RSA approach surpassed conventional color extraction techniques by achieving higher color accuracy of 96.84%, color contrast ratio of 7.50, inception score of 3.6, PSNR of 39.8&#xa0;dB, scalability of 97.7%, and lower error rate of 0.19. Overall, this research offers a significant contribution to computational design by providing a robust model that bridges aesthetics and algorithmic efficiency for diverse graphic design applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing graphic design through deep graph convolution and skill optimization: the DGC-RSA approach

  • 期刊论文

  • 发表日期：

    2025年8月21日

查看全文
  • Mingyi Sun,
  • Zhuyuan He,
  • Renyong Huang

摘要

Graphic design is a visual communication tool that communicates messages using elements like text, imagery, color, and layout. Among these elements, accurate color element extraction is essential to maintain consistency, optimize color contrasts, and create visually engaging designs. Therefore, this paper introduces a Deep Graph Convolutional based Random Skill Algorithm (DGC-RSA) that combines Deep Graph Convolutional Networks (DGCN) and RSA to optimize color feature analysis in graphic design. Initially, pixel-based color extraction is performed for color clustering and the Median-Cut algorithm is used to perform color quantification. Secondly, perceptual characteristics such as hue, saturation, and brightness are analyzed globally and locally to optimize emotional and communicative impact. The core innovation lies in the DGCN which models color clusters as graph nodes and spatial relationships as edges to derive meaningful color patterns, thereby enabling the framework to capture intricate contextual dependencies and enhance design harmony. Moreover, the hyperparameters of the DGCN are fine-tuned using an RSA algorithm that enables efficient exploration and exploitation of the search space to ensure convergence by adapting dynamically and avoiding local optima through randomness. The comprehensive evaluation is conducted by using an Image colorization dataset using diverse metrics including, color contrast ratio, color accuracy, inception score, error rate, PSNR, and scalability. The results revealed that the DGC-RSA approach surpassed conventional color extraction techniques by achieving higher color accuracy of 96.84%, color contrast ratio of 7.50, inception score of 3.6, PSNR of 39.8 dB, scalability of 97.7%, and lower error rate of 0.19. Overall, this research offers a significant contribution to computational design by providing a robust model that bridges aesthetics and algorithmic efficiency for diverse graphic design applications.