Edge-Preserving Single Image Super-Resolution

Qiang Zhou, Shifeng Chen, Jianzhuang Liu, and Xiaoou Tang


Abstract

This paper proposes a novel approach to single image superresolution. First, an image up-sampling scheme is proposed which takes the advantages of both bilateral filtering and mean shift image segmentation. Then we use a shock filter to enhance strong edges in the initial up-sampling result and obtain an intermediate high-resolution image. Finally, we enforce a reconstruction constraint on the highresolution image so that fine details can be inferred by back projection. Since strong edges in the intermediate result are enhanced, ringing artifacts can be suppressed in the back projection step. We compare our algorithm with several state-of-the-art image super-resolution algorithms. Qualitative and quantitative experimental results demonstrate that our approach performs the best.


 

 

Figure 1: Framework of our approach.

 

 

Figure 2: Comparison of different algorithms at 3 magnification on the mickey and Lena images. (a) Bicubic interpolation. (b) Sharpen bicubic (bicubic interpolation + unsharp mask in Adobe Photoshop). (c) Back projection. (d) Our approach.

 

 

Figure 3: Comparison with a reconstruction based method at 3 magnification on the Hat image. (a) Original image zoomed-in by nearest interpolation. (b) Bicubic interpolation. (c) Result by [13]. (d) Our approach.

 

 

Figure 4: Comparison with another reconstruction based method at 3 magnification on the Zebra image. (a) Original image zoomed-in by nearest interpolation. (b) Bicubic Interpolation. (c) Back Projection. (d) Result by [4]. (e) Our approach.

 

 

Figure 5: Comparison with the learning based method at 3 magnification using the Man image. (a) Original image zoomed-in by nearest interpolation. (b) Back projection. (c) Result by [4]. (d) Result by [21]. (e) Our approach.

 

 

Table 1: RMS values obtained by several methods. The best results are in bold.

 

 

Table 2: SSIM values obtained by several methods. The best results are in bold.

 

 

 

Ÿ  References:

Q. Zhou, S. Chen, J. Liu, and X. Tang, "Edge-Preserving Single Image Super-Resolution," ACM Multimedia (ACM-MM), 2011. [pdf]