Journal of Computer Applications ›› 2016, Vol. 36 ›› Issue (4): 1096-1099.DOI: 10.11772/j.issn.1001-9081.2016.04.1096

Previous Articles     Next Articles

Single image super-resolution via independently adjustable sparse coefficients

NI Hao1, RUAN Ruolin2, LIU Fanghua1, WANG Jianfeng3   

  1. 1. College of Electronic and Information Engineering, Hubei University of Science and Technology, Xianning Hubei 437100, China;
    2. College of Biomedical Engineering, Hubei University of Science and Technology, Xianning Hubei 437100, China;
    3. Center of Network Management, Hubei University of Science and Technology, Xianning Hubei 437100, China
  • Received:2015-07-30 Revised:2015-10-29 Online:2016-04-10 Published:2016-04-08
  • Supported by:
    This work is partially supported by the National Natural Science Foundation of China (61271256), the Team Plan Program of the Outstanding Young Science and Technology Innovation of Colleges and Universities in Hubei Province (T201513), the Natural Science Foundation of Hubei Province (2015CFB452), the Research Project of Education Department in Hubei Province (B2015080).

稀疏系数独立可调的单图超分辨率重建

倪浩1, 阮若林2, 刘芳华1, 王建峰3   

  1. 1. 湖北科技学院 电子与信息工程学院, 湖北 咸宁 437100;
    2. 湖北科技学院 生物医学工程学院, 湖北 咸宁 437100;
    3. 湖北科技学院 网络管理中心, 湖北 咸宁 437100
  • 通讯作者: 阮若林
  • 作者简介:倪浩(1981-),男,湖北麻城人,讲师,硕士,主要研究方向:机器学习、计算机视觉; 阮若林(1974-),男,湖北红安人,教授,博士,主要研究方向:视音频编码、图像处理; 刘芳华(1982-),女,湖北孝感人,讲师,硕士,主要研究方向:图像处理; 王建峰(1979-),男,湖北咸宁人,高级工程师,硕士,主要研究方向:计算机网络、图像处理。
  • 基金资助:
    国家自然科学基金资助项目(61271256);湖北省高等学校优秀中青年科技创新团队计划项目(T201513);湖北省自然科学基金资助项目(2015CFB452);湖北省教育厅科研计划指导性项目(B2015080)。

Abstract: The recovered image from the example-based super-resolution has sharp edges, but there are obvious artifacts. An improved super-resolution algorithm with independently adjustable sparse coefficients was proposed to eliminate the artifacts. In the dictionary training phase, the sparse coefficients in the high-dimensional space and the low-dimensional space of the image are different because of the known high-resolution training images and low-resolution ones. So the accurate high-resolution dictionary and the low-resolution one were generated separately via online dictionary learning algorithm. In the image reconstruction phase, the sparse coefficients in the two spaces were approximately the same because the input low-resolution image was known but the target high-resolution image was unknown. Different regularization parameters in the two phases were set to tune the corresponding sparse coefficients independently to get the best super-resolution results. According to the experiment results, the Peak Signal-to-Noise Ratio (PSNR) of the proposed method is 0.45 dB higher than that of sparse coding super-resolution in average, while the Structural SIMilarity (SSIM) is also 0.011 higher. The proposed algorithm eliminates the artifacts as well as recovers the edge sharpness and texture details effectively to promote the super-resolution results.

Key words: sparse coefficient, super-resolution reconstruction, online dictionary learning, single image

摘要: 针对基于学习的超分辨率重建图像边缘锐度较好但伪影较明显的问题,提出一种改进的稀疏系数独立可调的超分算法以消除伪影。由于字典训练阶段高分辨率图像和低分辨率图像均已知,认为高维图像空间和低维图像空间对应的稀疏系数不同,故此阶段运用在线字典学习方法分开训练生成较精确的高分字典和低分字典;而在图像重建阶段低分图像已知而高分图像未知,认为两空间的稀疏系数是近似相同的。通过在这两个阶段设置不同的正则化参数,可独立地调整相应的稀疏系数以获得最好的超分效果。实验结果表明,目标高分图像峰值信噪比(PSNR)相比稀疏编码超分方法平均提高了0.45 dB,同时结构相似性(SSIM)指标增加了0.011。超分图像有效地抑制了伪影,并能够较好地恢复图像边缘锐度和纹理细节,提升了超分效果。

关键词: 稀疏系数, 超分辨率重建, 在线字典学习, 单图

CLC Number: