Image inpainting based on scene transform and color transfer

Authors:
Hong Li;Song Wang;Wei Zhang;Min Wu
Affiliations:
School of Information Science and Engineering, Central South University, Changsha 410083, China;School of Information Science and Engineering, Central South University, Changsha 410083, China and School of Computer, XiangNan University, Chenzhou 423000, China;School of Architecture, Hunan University, Changsha 410082, China;School of Information Science and Engineering, Central South University, Changsha 410083, China
Venue:
Pattern Recognition Letters
Year:
2010

Citing 10
Cited 3

Color indexing

International Journal of Computer Vision
Image inpainting

Proceedings of the 27th annual conference on Computer graphics and interactive techniques
Color Transfer between Images

IEEE Computer Graphics and Applications
Graphcut textures: image and video synthesis using graph cuts

ACM SIGGRAPH 2003 Papers
Fragment-based image completion

ACM SIGGRAPH 2003 Papers
Poisson image editing

ACM SIGGRAPH 2003 Papers
Context-based vision system for place and object recognition

ICCV '03 Proceedings of the Ninth IEEE International Conference on Computer Vision - Volume 2
Scene completion using millions of photographs

ACM SIGGRAPH 2007 papers
Selective color transfer with multi-source images

Pattern Recognition Letters
Image quality assessment: from error visibility to structural similarity

IEEE Transactions on Image Processing

Image inpainting with salient structure completion and texture propagation

Pattern Recognition Letters
Accurate spatio-temporal reconstruction of missing data in dynamic scenes

Pattern Recognition Letters
Combining anisotropic diffusion, transport equation and texture synthesis for inpainting textured images

Pattern Recognition Letters

Quantified Score

Hi-index	0.10

Visualization

Abstract

In this paper, an image inpainting algorithm named stct-inpainting is proposed based on scene transform and color transfer. At first, a source image, which wears the most similar appearance to the damaged image, is searched from the image database by comparing with the features like texture, color and structural information. Then, scene filling in damaged region is completed by selecting appropriate object from the source image. After that, a cost function is employed to tackle the boundary line of the inpainted region and a color transfer algorithm is finally used to make the color appearance of the inpainted region be in harmony with that of its surroundings. Experimental results demonstrate that our developed algorithm is comparable with the state-of-art scene-completion algorithm proposed by Hays.