PDE image inpainting with texture synthesis based on damaged region classification
Authors:
- 1 January 2010
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 5, 447-450
Abstract
Image inpainting aims to restore lost information according to gray patterns from known area. PDE and texture synthesis are both widely applied in image inpainting. A new image inpainting algorithm combined those two kinds of methods are presented. Damaged region pixels are classified into texture pixels and structure pixels, which are inpainted with corresponding algorithms. Texture classification to the known image is also done to reduce the texture search region. The new algorithm require much less time than the classic inpainting algorithm.Keywords
This publication has 11 references indexed in Scilit:
- Variational denoising of partly textured images by spatially varying constraintsIEEE Transactions on Image Processing, 2006
- Image Decomposition into a Bounded Variation Component and an Oscillating ComponentJournal of Mathematical Imaging and Vision, 2005
- Region Filling and Object Removal by Exemplar-Based Image InpaintingIEEE Transactions on Image Processing, 2004
- Image restoration using multiresolution texture synthesis and image inpainting2004
- Object removal by exemplar-based inpainting2003
- Structure and texture filling-in of missing image blocks in wireless transmission and compression applicationsIEEE Transactions on Image Processing, 2003
- Curve evolution implementation of the Mumford-Shah functional for image segmentation, denoising, interpolation, and magnificationIEEE Transactions on Image Processing, 2001
- Image inpainting2000
- Texture synthesis by non-parametric sampling1999
- Is there any texture in the image?Pattern Recognition, 1996