YANG Xue-feng, CHENG Yao-yu
（School of Information and Communication Engineering, North University of China, Taiyuan 030051, China）
Abstract： Face hallucination or super-resolution is an inverse problem which is underdetermined, and the compressive sensing(CS) theory provides an effective way of seeking inverse problem solutions. In this paper, a novel compressive sensing based face hallucination method is presented, which is comprised of three steps： dictionary learning、 sparse coding and solving maximum a posteriori (MAP) formulation. In the first step, the K-SVD dictionary learning algorithm is adopted to obtain a dictionary which can sparsely represent high resolution(HR) face image patches. In the second step, we seek the sparsest representation for each low-resolution (LR) face image paches input using the learned dictionary, super resolution image blocks are obtained from the sparsest coefficients and dictionaries, which then are assembled into super-resolution (SR) image. Finally, MAP formulation is introduced to satisfy the consistency restrictive condition and obtain the higher quality HR images. The experimental results demonstrate that our approach can achieve better super-resolution faces compared with other state-of-the-art method.
Key words： face image; super-resolution image; face hallucination; compressive sensing (CS)
CLD number： TP391.41Document code： A
Article ID： 1674-8042（2016）02-0149-06 doi： 10.3969/j.issn.1674-8042.2016.02.009
References
［1］Farsiu S, Robinson M D, Elad M, et al. Fast and robust multiframe super-resolution. IEEE Transactions on Image Processing, 2004, 13(10)： 1327-1344.
［2］SHEN Huan-feng, ZHANG Liang-pei, HUANG Bo, et al. A map approach for joint motion estimation, segmentation, and super resolution. IEEE Transactions on Image Processing, 2007, 16(2)： 479-490.
［3］ZHANG Kai-bing, TAO Da-cheng, GAO Xin-bo, et al. Learning multiple linear mappings for efficient single image super-resolution. IEEE Transactions on Image Processing, 2015, 24(3)： 846-861.
［4］Ni K S, Nguyen T Q. Image superresolution using support vector regression. IEEE Transactions on Image Processing, 2007, 16(6)： 1596-1610.
［5］Baker S, Kanade T. Limits on super-resolution and how to break them. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002, 24(9)： 1167-1183.
［6］ZHUANG Yue-ting, ZHANG Jian, WU Fei. Hallucinating faces： LPH super-resolution and neighbor reconstruction for residue compensation. Pattern Recognition, 2007, 40： 3178-3194.
［7］YANG Jian-chao, Wright J, Huang T S, et al. Image super-resolution via sparse representation. IEEE Transactions on Image Processing, 2010, 19(11)： 2861-2873.
［8］WANG Nan-nan, TAO Da-cheng, GAO Xin-bo, et al. A comprehensive survey to face hallucination. International Journal of Computer Vision, 2014, 106(1)： 9-30.
［9］MA Jian-wei, le Dimet F X. Deblurring from highly incomplete measurements for remote sensing. IEEE Transactions on Geoscience and Remote Sensing, 2009, 47(3)： 792-802.
［10］Elad M, Aharon M. Image denoising via sparse and redundant representations over learned dictionaries. IEEE Transactions on Image Processing, 2006, 15(12)： 3736-3745.
［11］Kulkarni N, Nagesh P, Gowda R, et al. Understanding compressive sensing and sparse representation-based super-resolution. IEEE Transactions on Circuits & Systems for Video Technology, 2012, 22(5)： 778-789.
［12］Candès E J, Romberg J, Tao T. Robust uncertainty principles： exact signal reconstruction from highly incomplete frequency information. IEEE Transactions on Information Theory, 2006, 52(2)： 489-509.
［13］Donoho D L. Compressed sensing. IEEE Transactions on Information Theory, 2006, 52(4)： 1289-1306.
［14］Baraniuk R G. Compressive sensing. IEEE Signal Processing Magazine, 2007, 24(4)： 118-124.
［15］Romberg J. Imaging via compressive sampling. IEEE Signal Processing Magazine, 2008, 25(2)： 14-20.
［16］Rauhut H, Schnass K, Vandergheynst P. Compressed sensing and redundant dictionaries. IEEE Transactions on Information Theory, 2008, 54(5)： 2210-2219.
［17］Devore R A. Deterministic constructions of compressed sensing matrices. Journal of Complexity, 2007, 23： 918-925.
［18］Aharon M, Elad M, Bruckstein A. K-SVD： An algorithm for designing of overcomplete dictionaries for sparse representation. IEEE Transactions on Signal Process, 2006, 54(11)： 4311-4322.
基于压缩感知的人脸超分辨技术
杨学峰， 程耀瑜
（中北大学 信息与通信工程学院 ， 山西 太原 030051）
摘要： 人脸超分辨是一个欠定的求逆问题， 压缩感知理论提供了一种有效的逆问题求解方法。 本文提出了一种基于压缩感知的人脸超分辨方法， 该方法包含三个步骤： 字典学习、 稀疏编码和求解全局MAP 方程。 第一步， 使用K-SVD算法获得能稀疏表示高分辨率人脸图像的字典； 第二步利用训练好的字典， 求解输入低分辨率图像块的最稀疏表达， 由稀疏系数和稀疏字典求得超分辨图像块， 组装成超分辨图像； 最后， 为了满足全局一致性限制条件和改进超分辨效果， 对完整图像求解MAP方程。 实验结果表明， 与其它同类人脸超分辨方法相比， 本文方法得到了更好的超分辨效果。
关键词： 人脸图像； 超分辨； 虚幻脸； 压缩感知
引用格式： YANG Xue-feng, CHENG Yao-yu. Face hallucination via compressive sensing. Journal of Measurement Science and Instrumentation, 2016, 7(2)： 149-154. ［doi： 10.3969／j.issn.1674-8042.2016.02.009］
[full text view]
