基于数字报历史优秀版面的样式智能生成与微调

• 作者： 陶颖 ¹ 程雨夏 ¹ 曾振宇 ¹ 庄跃辉 ² 张艺馨 ¹ 何兴臻 ¹
• 作者单位：
  1.  杭州电子科技大学计算机学院 杭州 310000
  2.  浙江方正传媒技术研究院，金华 321000
• 通讯作者： 程雨夏  Email:yxcheng@hdu.edu.cn
• 提交时间：2022-07-14

摘要: 目的 报纸一直是传播知识的重要载体，本文方法为实现经济、高效的报纸排版工作。 方法 首先根据历史优秀版面训练概率模型来推断电子报版面的样式，并结合固定布局约束和用户约束保证样式有效，同时构建美学设计原理的量化方法进一步实现样式微调。 结果 通过定性和定量评估，表明由本文模型推断出的样式参数精确度良好，且满足用户一定的需求。 局限 本文方法暂时只支持单页电子报的自动生成，然而报纸排版多由多个版面组成，故未来的工作需要对报纸内容进行分页操作。 结论 本文方法可以自动生成满足视觉美观性、层次性和可读性的报纸。

Abstract: Objective Newspapers have always been an important carrier of knowledge dissemination. This method is to achieve economic and efficient newspaper typesetting. Methods First,we infer the style of newspaper layout according to the historical excellent layouts training probability model, and combine the fixed layout constraints and user constraints to ensure that the style is effective. At the same time,we build a quantitative method of aesthetic design principles to further realize style fine-tuning. Results Through qualitative and quantitative evaluation, it shows that the style parameters inferred from the model in this paper are accurate and meet the needs of users. Limitations The method only supports the automatic generation of single page electronic newspapers temporarily. However, newspaper layout is mostly composed of multiple pages, so the future work needs to page the newspaper content. Conclusions This method can automatically generate newspapers that meet the requirements of visual aesthetics, hierarchy and readability.

• 来自： 陶颖
• 分类： 计算机科学 >> 计算机应用技术
• 投稿状态： 未投稿
• 引用： ChinaXiv:202207.00113 (或此版本 ChinaXiv:202207.00113V2)
  doi:10.12074/202207.00113
  CSTR:32003.36.ChinaXiv.202207.00113.V2
• 推荐引用方式： 陶颖,程雨夏,曾振宇,庄跃辉,张艺馨,何兴臻.(2022).基于数字报历史优秀版面的样式智能生成与微调.中国科学院科技论文预发布平台.[ChinaXiv:202207.00113] (点此复制)