基于多级隐空间信息约束的噪声人脸超分辨率算法

Noisy face super-resolution method based on multi-level latent space information constraints

为了实现强噪声和模糊干扰下的低清人脸图像重建,提出一种基于多级隐空间信息约束的噪声人脸超分辨率算法.首先设计一个用于人脸有效信息提取的特征蒸馏网络,并通过统计性抗干扰模型和隐空间特征对比算法移除噪声等无效信息,构建一个具有高噪声鲁棒性的人脸信息提取模型;然后,设计人脸重建网络,该网络利用提取的人脸特征重建高清人脸图像;最后,通过人脸身份嵌入模型和离散小波变换模型,分别从超球面身份度量空间和小波域进一步对重建人脸的身份信息和空间结构进行约束.实验结果表明,所提出的算法不仅能够有效去除高噪声环境下的人脸噪声,而且还能有效提升人脸图像分辨率,获得更高的峰值信噪比(peak signal-to-noise ratio, PSNR)和结构相似度(structural similarity index, SSIM),具有较好的实用性.

To super-resolve low-resolution(LR)face image suffering from strong noise and fuzzy interference,we propose a noisy face super-resolution(FSR)method based on multi-level latent space information constraints.Firstly,we design a feature distillation network to extract effective facial information,which exploits a statistical anti-interference model and a latent contrast algorithm to remove invalid information such as noise.Then,we design a face reconstruction network,which utilizes the extracted face features to reconstruct high-resolution(HR)face images.Finally,we deploy a face identity embedding model and a discrete wavelet transform model to further supervise the reconstruction of identity information and spatial structure from the hypersphere identity metric space and wavelet domain respectively.The experimental results show that the proposed method not only removes the noise from face in the high noise environment,but also improves the resolution of the face image effectively,which obtains higher PSNR,SSIM,and good practicability.