《计算机应用研究》|Application Research of Computers

基于改进CycleGAN的视频监控人脸超分辨率恢复算法

Improved video surveillance face super-resolution recovery algorithm based on CycleGAN

免费全文下载 (已被下载 次)  
获取PDF全文
作者 陈贵强,何军,罗顺茺
机构 四川大学 计算机学院,成都 610065
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2021)10-052-3172-05
DOI 10.19734/j.issn.1001-3695.2020.11.0434
摘要 针对有监督超分辨率算法训练过程需要大量成对图像、处理真实低分辨率图像视觉恢复效果差等问题,提出了一种基于改进CycleGAN的半监督算法Cycle-SRNet。首先,利用退化模型获得与真实低分辨率人脸相似的图像,用于训练网络参数;其次,通过重建模型恢复出具有真实效果的高分辨率人脸图像;最后引入感知损失函数保持人脸结构相似性,以更好地恢复面部特征。实验结果表明,该算法不需要成对的图像进行网络训练,在视觉效果上能够将模糊的视频监控低分辨率人脸图像恢复成清晰可辨的人脸图像,在FID、PSNR和SSIM指标上超越了SRCNN、SRGAN、CinCGAN等方法。
关键词 单幅图像超分辨率恢复; 生成对抗网络; CycleGAN; 半监督学习; 人脸超分辨率
基金项目 国家自然科学基金资助项目(U1836103)
四川省科技重点研发项目(18ZDYF2039)
四川省重大科技专项(2017GZDZX0002)
本文URL http://www.arocmag.com/article/01-2021-10-052.html
英文标题 Improved video surveillance face super-resolution recovery algorithm based on CycleGAN
作者英文名 Chen Guiqiang, He Jun, Luo Shunchong
机构英文名 College of Computer Science,Sichuan University,Chengdu 610065,China
英文摘要 Supervised super-resolution algorithms require plentiful paired-image for networks training, resulting in poor visual recovery effects when processing real-world low-resolution images. To solve this problem, this paper proposed an improved semi-supervised algorithm based on CycleGAN, donated as Cycle-SRNet. Firstly the degradation model generated images similar to real world low-resolution faces for network training, and then the reconstruction model restored high resolution face image. Furthermore, this paper introduced perceptual loss function to ensure the structural similarity of the faces so as to recover the facial features better. Experimental results show that the algorithm can restore fuzzy low-resolution video surveillance face images into recognizable face images, and it exceeds SRCNN, SRGAN and CinCGAN in terms of FID, PSNR and SSIM indicators.
英文关键词 single image super-resolution(SISR); GAN; CycleGAN; semi-supervised; face super-resolution
参考文献 查看稿件参考文献
 
收稿日期 2020/11/24
修回日期 2021/1/12
页码 3172-3176
中图分类号 TP399
文献标志码 A