基于独立分量分析的人脸超分辨率重建

提出一种基于独立分量分析(ICA)的人脸超分辨率重建算法。该算法利用ICA从高分辨率训练图像中提取出独立分量，并对ICA系数进行先验估计。对于给定的低分辨率图像，结合最大后验概率估计求出ICA系数，进行ICA反变换得到高分辨率图像的近似估计，并利用局部结构张量对图像进行精化处理得到重建图像。仿真结果表明，该算法在实现人脸超分辨率重建的同时保持了人脸整体结构特征，且对光照、表情、姿态等具有一定的鲁棒性，将重建结果用于人脸辨识，有效提高了辨识效率。

Face Super-resolution Reconstruction Based on ICA

This paper proposes a face super-resolution algorithm based on Independent Components Analysis（ICA）. It obtains ICs from offiine training high-resolution face images by ICA. The prior of ICA coefficients are estimated by performing PCA on training images. For a low-resolution image, the high-resolution image is reconstructed by the linear combination of the ICs where the weight coefficients are obtained by Maximum A Posteriori probability（MAP）. A structure-tensor-based filter is proposed to refine the estimated image. Experimental results demonstrate that the algorithm is robust to various pose, expressions and lighting conditions. And the hallucination results preserve both the global structure and the high spatial-frequency information better such as sharp edges and high contrast. After that, reconstruction results are used for face recognition which improves the recognition rate.