| 基于条件生成对抗网络生成人体肖像线稿 |
| |
| 引用本文： | 刘鹏飞,杨威.基于条件生成对抗网络生成人体肖像线稿[J].计算机应用与软件,2020,37(1):237-241,255. |
| |
| 作者姓名： | 刘鹏飞 杨威 |
| |
| 作者单位： | 中国科学技术大学苏州研究院 江苏 苏州215000;中国科学技术大学苏州研究院 江苏 苏州215000 |
| |
| 摘 要： | 将深度学习用于图像边缘提取,获取线稿图。常见的图像边缘获取算法存在两个缺陷:只考虑到了图像边缘的一阶、二阶数据特征,并未发掘其高阶数据特征及其他隐含特征;没有设定合理的阈值,而阈值直接影响着检测效果。这两个因素使得到的图像线稿图效果不佳。鉴于此,提出基于条件对抗网络生成人体肖像线稿图的方案,并在最终的实验中取得了良好效果。
|
| 关 键 词： | 条件生成对抗网络 人体肖像 线稿 |
GENERATING HUMAN PORTRAIT LINE DRAWING BASED ON CONDITIONAL GENERATIVE ADVERSARIAL NETWORKS |
| |
| Liu Pengfei,Yang Wei.GENERATING HUMAN PORTRAIT LINE DRAWING BASED ON CONDITIONAL GENERATIVE ADVERSARIAL NETWORKS[J].Computer Applications and Software,2020,37(1):237-241,255. |
| |
| Authors: | Liu Pengfei Yang Wei |
| |
| Affiliation: | (Suzhou Research Institute,University of Science and Technology of China,Suzhou 215000,Jiangsu,China) |
| |
| Abstract: | Deep learning is used for image edge extraction to obtain line drawing.Common image edge acquisition algorithms have two defects:only take into account the first-order and second-order data features of the image edge,and do not explore its high-order data features and other hidden features;these methods are needed to set reasonable thresholds,and thresholds directly affect the detection effect.These two factors make the result of the image line drawing not good.In view of this,this paper proposes a scheme of generating a human portrait line drawing based on a conditional confrontation network,and achieves good results in the final experiment. |
| |
| Keywords: | Conditional generative adversarial networks Human portrait Line drawing |
| 本文献已被 维普 万方数据 等数据库收录！ |
|
