Abstract | ||
---|---|---|
We present a robust algorithm for spatial recovery of missing region in images. The algorithm consists of two stages: sparse modeling and patch based refinement. We note that a model based image recovery might not be able to reconstruct the richness or details in a signal unless the signal truly fits that model. We show that the reconstruction using a sparse model provides enough information about the inherent features present in the unknown area, using which, a patch based refinement process can replicate the structure and the natural texture from the surrounding available samples. The developed algorithm is tested on a variety of image characteristics. Significant objective and subjective gains are observed compared to the state-of-the-art. |
Year | DOI | Venue |
---|---|---|
2010 | 10.1109/ICASSP.2010.5494974 | Acoustics Speech and Signal Processing |
Keywords | Field | DocType |
image reconstruction,image texture,refinement calculus,sparse matrices,image recovery,missing region,natural texture,patch based refinement,sparse modeling,sparse reconstruction,spatial recovery,structure replication,texture refinement,Error concealment,Image recovery,Sparse reconstruction,Texture refinement | Iterative reconstruction,Computer vision,Pattern recognition,Refinement calculus,Image texture,Sparse model,Computer science,Image segmentation,Artificial intelligence,Image recovery,Replicate,Sparse matrix | Conference |
ISSN | ISBN | Citations |
1520-6149 E-ISBN : 978-1-4244-4296-6 | 978-1-4244-4296-6 | 4 |
PageRank | References | Authors |
0.72 | 7 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Haricharan Lakshman | 1 | 328 | 30.58 |
Martin Köppel | 2 | 4 | 0.72 |
Patrick Ndjiki-Nya | 3 | 4 | 0.72 |
Thomas Wiegand | 4 | 3348 | 279.51 |