低质量监控图像鲁棒性人脸超分辨率算法

Robust Super-resolution Algorithm for Low-quality Surveillance Face Images

由于人对图像结构信息的理解对于像素值的噪声干扰具有极强的鲁棒功能,为了增强传统算法针对低质量监控图像的鲁棒性,提出一种基于人工形状语义模型的人脸超分辨率算法.该算法将形状描述成一系列面部特征点的组合,通过人工获取人脸图像形状语义信息,利用形状样本库构建超分辨率代价函数的正则约束项;将图像与形状的系数相关性用于统一重建误差项与形状正则项的变量,并将最速下降法用于优化求解.仿真和实际图像实验结果都表明,在主客观质量上,文中算法的性能都优于传统算法.

Human understanding with image semantic information,especially structural information,is robust to the degraded pixel values.In order to enhance the robustness of traditional methods to low-quality surveillance images,we propose a face super-resolution approach using shape semantic model.This method describes the facial shape as a series of fiducial points on facial image.And shape semantic information of input image is obtained manually.Then a shape semantic regularization is added to the original objective function.According to the correlation of coefficients of image and shape,the variables of reconstruction fidelity term and shape regularization item are unified.And the steepest descent method is used to obtain the unified coefficient.Experimental results of simulation and real images indicate that the proposed method outperforms the traditional schemes significantly both in subjective and objective qualities.