低秩矩阵恢复的超分辨图像重建算法
|
摘要
超分辨图像重建是提高图像质量的重要技术,为了提高图像重建的精度,以获得高质量的图像,针对当前图像重建算法存在缺陷,提出了基于低秩矩阵恢复的超分辨图像重建算法。首先根据低分辨率图像和高分辨率图像间的关系,选择训练样本,并将图像划分为多个子块,然后采用低秩矩阵恢复算法对子块进行学习,根据块与块间的相关性找到低分辨和高分辨图像子块重建的权值,实现图像超分辨率重建,最后在Matlab 2012平台上进行了图像超分辨率重建实验,结果表明,该算法提高了图像重建的精度,保留了边缘等细节信息,重建速度可以满足图像处理的实时性要求,而且重建效果要优于当前经典的图像重建算法。
Super resolution image reconstruction is an important technique to improve image quality. In order to improve the accuracy of image reconstruction and obtain high quality images, aiming at the defects of the current image reconstruction algorithms, a novel super resolution image reconstruction algorithm is proposed by using low rank matrix recovery. Firstly, training samples are divided into several sub blocks according to the relationship between the low resolution image and the high resolution image, and then the low rank matrix recovery algorithm is used to learn sub blocks, the reconstruction weight of low resolution and high resolution image sub blocks is found according to the correlation between blocks, and super resolution images are reconstructed, finally, the image super resolution reconstruction experiment is carried out on the Matlab 2012 platform. The results show that the proposed algorithm can effectively improve image reconstruction accuracy, and can keep edges and other details, the image reconstruction speed can meet the real-time requirements of image processing, and the reconstruction effect is better than classical algorithms.
Super resolution image reconstruction is an important technique to improve image quality. In order to improve the accuracy of image reconstruction and obtain high quality images, aiming at the defects of the current image reconstruction algorithms, a novel super resolution image reconstruction algorithm is proposed by using low rank matrix recovery. Firstly, training samples are divided into several sub blocks according to the relationship between the low resolution image and the high resolution image, and then the low rank matrix recovery algorithm is used to learn sub blocks, the reconstruction weight of low resolution and high resolution image sub blocks is found according to the correlation between blocks, and super resolution images are reconstructed, finally, the image super resolution reconstruction experiment is carried out on the Matlab 2012 platform. The results show that the proposed algorithm can effectively improve image reconstruction accuracy, and can keep edges and other details, the image reconstruction speed can meet the real-time requirements of image processing, and the reconstruction effect is better than classical algorithms.
引文
[1]张洪艳,沈焕锋,张良培等.一种保边缘影像超分辨率重建方法[J].中国图象图形学报,2009,14(11):2255-2261.Zhang H Y,Shen H F,Zhang L P,et al.An Edge-preserving Image Super-resolution Reconstruction Method[J].Journal of Image and Graphics,2009,14(11):2255-2261.
[2]蒋建国,董艳春.边缘增强型非局部模型超分辨率重建算法[J].中国图象图形学报,2011,16(10):1772-1778.Jang J G,Dong Y Cn.Edge-enhanced Nonlocal Model Super-resolution Reconstruction[J].Journal of Image and Graphics,2011,16(10):1772-1778.
[3]安耀祖,陆耀,赵红.一种自适应正则化的图像超分辨率算法[J].自动化学报,2012,38(4):601-608.An Y Z,Lu Y,Zhao H.An Adaptive-regularized Image Super-resolution[J].Acta Automatica Sinica,2012,38(4):601-608.
[4]肖创柏.一种基于MAP的超分辨率图像重建的快速算法[J].计算机研究与发展,2009,46(5):872-880.Xiao C B.A Novel Fast Algorithm for MAP Super-Resolution Image Reconstruction[J].Journal of Computer Research and Development,2009,46(5):872-880.
[5]Takeda H,Milanfar P,Protter M,et al.Super-resolution Without Explicit Sub-pixel Motion Estimation[J].IEEE Transactions on Image Processing,2009,18(9):1958-1975.
[6]Yang J C,Wrigh J,Huang T S,Ma Y.Image Super-resolution Via Sparse Representation[J].IEEE Trans on Image Process,2010,19(11):2861-2873.
[7]杨欣,费树岷,周大可.基于MAP的自适应图像配准及超分辨率重建[J].仪器仪表学报,2011,20(8):1771-1775.Yang X,Fei S M,Zhou D Ke.Self-adapting Weighted Technology for Simultaneous Image Registration and Super-resolution Reconstruction Based on MAP[J].Chinese Journal of Scientific Instrument,2011,20(8):1771-1775.
[8]Gao X B,Zhang K B,Tao D C,etal.Joint Learning for Single-image Super-resolution Via a Coupled Constraint[J].IEEE Trans.On Image Process,2012,21(2):469–480.
[9]李展,张庆丰,孟小华等.多分辨率图像序列的超分辨率重建[J].自动化学报,2012,38(11):1804-1814.Li Z,Zhang Q F,Meng X H,et al.Super-resolution Reconstruction for Multi-resolution Image Sequence[J].Acta Automatica Sinica,2012,38(11):1804-1814.
[10]樊博,杨晓梅,胡学姝.基于压缩感知的超分辨图像重建[J].计算机应用,2013,33(2):480-483.Fan B,Yang X M,Hu X S.Super Resolution Image Reconstruction Based on Compressed Sensing[J]Computer Application,2013,33(2):480-483.
[11]谭台哲,尚鹏.纹理约束的局部自学习单幅图像超分辨率算法[J].电视技术,2014,38(13):24-28.Tan T Z,Shang P.Texture-constrained Single Image Super-resolution From Local Self-examples[J].Tv Engineering,2014,38(13):24-28.
[12]范开乾,胡访宇.基于支持向量回归的图像超分辨率重建算法[J].电子技术,2014,4(2):4-7.Fan K Q,Hu F Y.An Image Super Resolution Reconstruction Algorithm Based on Support Vector Regression[J].Electronic Technology,2014,4(2):4-7.
[13]禹晶,苏开娜,肖创柏.一种改善超分辨率图像重建中边缘质量的方法[J].自动化学报,2007,33(6):577-582.Yu J,Su K N,Xiao C B.Edge Artifact Reduction for Super-resolution Image Reconstruction[J].Acta Automatica Sinica,2007,33(6):577-582.
[14]陈晓璇,齐春.基于低秩矩阵恢复和联合学习的图像超分辨率重建[J].计算机学报,2014,37(6):1372-1379.Chen X X,Qi C.Single-Image Super-Resolution via Low-Rank Matrix Recovery and Joint Learning[J].Chinese Journal of Computers,2014,37(6):1372-1379.
[2]蒋建国,董艳春.边缘增强型非局部模型超分辨率重建算法[J].中国图象图形学报,2011,16(10):1772-1778.Jang J G,Dong Y Cn.Edge-enhanced Nonlocal Model Super-resolution Reconstruction[J].Journal of Image and Graphics,2011,16(10):1772-1778.
[3]安耀祖,陆耀,赵红.一种自适应正则化的图像超分辨率算法[J].自动化学报,2012,38(4):601-608.An Y Z,Lu Y,Zhao H.An Adaptive-regularized Image Super-resolution[J].Acta Automatica Sinica,2012,38(4):601-608.
[4]肖创柏.一种基于MAP的超分辨率图像重建的快速算法[J].计算机研究与发展,2009,46(5):872-880.Xiao C B.A Novel Fast Algorithm for MAP Super-Resolution Image Reconstruction[J].Journal of Computer Research and Development,2009,46(5):872-880.
[5]Takeda H,Milanfar P,Protter M,et al.Super-resolution Without Explicit Sub-pixel Motion Estimation[J].IEEE Transactions on Image Processing,2009,18(9):1958-1975.
[6]Yang J C,Wrigh J,Huang T S,Ma Y.Image Super-resolution Via Sparse Representation[J].IEEE Trans on Image Process,2010,19(11):2861-2873.
[7]杨欣,费树岷,周大可.基于MAP的自适应图像配准及超分辨率重建[J].仪器仪表学报,2011,20(8):1771-1775.Yang X,Fei S M,Zhou D Ke.Self-adapting Weighted Technology for Simultaneous Image Registration and Super-resolution Reconstruction Based on MAP[J].Chinese Journal of Scientific Instrument,2011,20(8):1771-1775.
[8]Gao X B,Zhang K B,Tao D C,etal.Joint Learning for Single-image Super-resolution Via a Coupled Constraint[J].IEEE Trans.On Image Process,2012,21(2):469–480.
[9]李展,张庆丰,孟小华等.多分辨率图像序列的超分辨率重建[J].自动化学报,2012,38(11):1804-1814.Li Z,Zhang Q F,Meng X H,et al.Super-resolution Reconstruction for Multi-resolution Image Sequence[J].Acta Automatica Sinica,2012,38(11):1804-1814.
[10]樊博,杨晓梅,胡学姝.基于压缩感知的超分辨图像重建[J].计算机应用,2013,33(2):480-483.Fan B,Yang X M,Hu X S.Super Resolution Image Reconstruction Based on Compressed Sensing[J]Computer Application,2013,33(2):480-483.
[11]谭台哲,尚鹏.纹理约束的局部自学习单幅图像超分辨率算法[J].电视技术,2014,38(13):24-28.Tan T Z,Shang P.Texture-constrained Single Image Super-resolution From Local Self-examples[J].Tv Engineering,2014,38(13):24-28.
[12]范开乾,胡访宇.基于支持向量回归的图像超分辨率重建算法[J].电子技术,2014,4(2):4-7.Fan K Q,Hu F Y.An Image Super Resolution Reconstruction Algorithm Based on Support Vector Regression[J].Electronic Technology,2014,4(2):4-7.
[13]禹晶,苏开娜,肖创柏.一种改善超分辨率图像重建中边缘质量的方法[J].自动化学报,2007,33(6):577-582.Yu J,Su K N,Xiao C B.Edge Artifact Reduction for Super-resolution Image Reconstruction[J].Acta Automatica Sinica,2007,33(6):577-582.
[14]陈晓璇,齐春.基于低秩矩阵恢复和联合学习的图像超分辨率重建[J].计算机学报,2014,37(6):1372-1379.Chen X X,Qi C.Single-Image Super-Resolution via Low-Rank Matrix Recovery and Joint Learning[J].Chinese Journal of Computers,2014,37(6):1372-1379.