arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2011.13611cs.CV
收藏

频域图像变换:更真实,更好的身份保护

Frequency Domain Image Translation: More Photo-realistic, Better Identity-preserving

University of Wisconsin-Madison(威斯康星大学麦迪逊分校);SenseTime Group Ltd.(商汤集团有限公司);Kwai Inc.(快手公司);Beihang University(北京航空航天大学);Tsinghua University(清华大学)

Mu Cai, Hong Zhang, Huijuan Huang, Qichuan Geng, Yixuan Li, Gao Huang

摘要:

基于GAN的方法彻底改变了图像到图像的转换。然而,现有的方法缺乏保持源域的身份的能力。因此,合成的图像往往会过度适应参考域,失去重要的结构特征,并遭受次优的视觉质量。为了解决这些挑战,我们提出了一种新的频域图像转换(FDIT)框架,利用频率信息来增强图像生成过程。我们的关键思想是将图像分解为低频和高频分量,其中高频特征捕获与身份类似的对象结构。我们的训练目标有利于在像素空间和傅立叶频谱空间中保存频率信息。我们广泛评估FDIT在五个大规模数据集和多个任务,包括图像转换和GAN反演。大量的实验和烧蚀表明,FDIT有效地保持了源图像的身份,并产生照片般逼真的图像。FDIT建立了最先进的性能,与以前的最佳方法相比,平均FID分数降低了5.6%。

英文摘要:

Image-to-image translation has been revolutionized with GAN-based methods. However, existing methods lack the ability to preserve the identity of the source domain. As a result, synthesized images can often over-adapt to the reference domain, losing important structural characteristics and suffering from suboptimal visual quality. To solve these challenges, we propose a novel frequency domain image translation (FDIT) framework, exploiting frequency information for enhancing the image generation process. Our key idea is to decompose the image into low-frequency and high-frequency components, where the high-frequency feature captures object structure akin to the identity. Our training objective facilitates the preservation of frequency information in both pixel space and Fourier spectral space. We broadly evaluate FDIT across five large-scale datasets and multiple tasks including image translation and GAN inversion. Extensive experiments and ablations show that FDIT effectively preserves the identity of the source image, and produces photo-realistic images. FDIT establishes state-of-the-art performance, reducing the average FID score by 5.6% compared to the previous best method.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑