摘要
针对基于非局部稀疏表示的单幅图像超分辨率重建方法易丢失图像块之间的差异,造成重建图像出现过度平滑的问题,提出一种基于非局部Laplacian稀疏表示的重建方法。在利用非局部正则项对相似图像块进行约束的同时,引入相似保护项对图像块间的差异性进行约束,提高重建图像的质量。实验结果表明,该方法与其它算法相比在主观视觉效果上取得明显改进,在客观评价指标上明显提高。
The single-image super-resolution methods via non-local sparse representation easily lead to the problem that the re-constructed images are over-smoothed because of losing the difference among the image patches to be reconstructed. To overcome this problem, a reconstruct method via non-local Laplacian sparse representation was proposed. A non-local regularization term was utilized to constrain the similar image patches, and at the same time, similarity protection term was introduced to constrain the difference among the image patches, thus improving the quality of the reconstructed image. Experimental results demonstrate that, compared with the other algorithms, the proposed approach not only achieves significant improvements on the subjective vision, but also increases significantly on the objective evaluation.
出处
《计算机工程与设计》
北大核心
2015年第12期3302-3305,3316,共5页
Computer Engineering and Design
基金
国家自然科学基金项目(31360277)
河南省教育厅科学技术研究重点基金项目(12B520011)
关键词
超分辨率
稀疏表示
非局部
差异性
相似保护项
super-resolution
sparse representation
non-local
difference
similarity protection term