上海大学学报(自然科学版) ›› 2018, Vol. 24 ›› Issue (2): 314-321.doi: 10.12066/j.issn.1007-2861.1780

• 研究论文 • 上一篇    下一篇

基于混合梯度最小化的铅笔画生成

王权1,2, 胡越黎1,2,3(), 燕明1,2, 胡云生1,2, 陈朋1,2   

  1. 1. 上海大学 机电工程与自动化学院, 上海 200444
    2. 上海市电站自动化技术重点实验室, 上海 200072
    3. 上海大学 微电子研究与开发中心, 上海 200072
  • 收稿日期:2016-05-01 出版日期:2018-04-30 发布日期:2018-05-07
  • 通讯作者: 胡越黎 E-mail:huyueli@shu.edu.cn

Pencil drawing production based on hybrid gradient minimization

WANG Quan1,2, HU Yueli1,2,3(), YAN Ming1,2, HU Yunsheng1,2, CHEN Peng1,2   

  1. 1. School of Mechatronic Engineering and Automation, Shanghai University, Shanghai 200444, China
    2. Shanghai Key Laboratory of Power Station Automation Technology, Shanghai 200072, China
    3. Microelectronic Research and Development Center, Shanghai University, Shanghai 200072, China
  • Received:2016-05-01 Online:2018-04-30 Published:2018-05-07
  • Contact: HU Yueli E-mail:huyueli@shu.edu.cn

摘要:

利用计算机模拟铅笔画生成的算法主要存在两个问题: 其一是画面细节易丢失, 边缘重点线条不够突出; 二是笔画单一不能很好地展示原画中的层次信息. 针对以上问题, 提出了一种基于混合梯度最小化的铅笔画生成算法, 通过结合 $L_{0}$ 梯度最小化和 $L_{1}$ 范数约束, 既突出显示了图像边缘和重点线条, 又不影响图像的整体锐度, 同时可与原图像结构保持高度一致, 充分展现了图像的层次性和丰富的细节信息. 实验结果表明, 本算法能够生成质量更高、真实感更强的的铅笔画效果, 相比于已有的其他算法具有一定的优越性.

关键词: 铅笔画, 非真实感渲染, $L_{0}$ 梯度最小化, 色调渲染

Abstract:

Algorithms for creating pencil drawing effect from an image have two major problems: details are often missing so that important edge lines are not clearly shown, and lines are monotonous without hierarchical information. To solve these problems, this paper proposes a pencil drawing generation algorithm based on hybrid gradient minimization. By combining $L_0$ gradient minimization and $L_1$ norm, prominent edges are maintained and enhanced without affecting the overall sharpness. This way, rich and hierarchical details in the image are fully retained. Experimental results indicate that the proposed method can generate high quality and more realistic pencil drawing effect. It is superior to several other existing method.

Key words: pencil drawing, non-photorealistic rendering, $L_0$ gradient minimization, tone rendering

中图分类号: