Title :
Adaptive total variation deringing method for image interpolation
Author :
Krylov, Andrey ; Nasonov, Andrey
Author_Institution :
Fac. of Comput. Math. & Cybern., Moscow Lomonosov State Univ., Moscow
Abstract :
This paper presents a new adaptive post-processing algorithm for ringing artifact reduction after image interpolation (up sampling). The algorithm is based on the concept of total variation (TV) for ringing control. It uses known TV of the blocks of the low-resolution image. Conditional gradient, subgradient and projection subgradient methods for this algorithm are considered and analyzed. A test set of 181300 overlapping 11times11 blocks of real images was used for local algorithm optimization and analysis. Local conditional gradient method shows the best objective and subjective results.
Keywords :
gradient methods; image resolution; interpolation; adaptive postprocessing algorithm; adaptive total variation deringing; image interpolation; local conditional gradient method; low-resolution image; projection subgradient method; ringing artifact reduction; Algorithm design and analysis; Compression algorithms; Filtering algorithms; Fractals; Image edge detection; Image processing; Interpolation; Mathematics; TV; Transform coding; deringing; image enhancement; image interpolation; total variation;
Conference_Titel :
Image Processing, 2008. ICIP 2008. 15th IEEE International Conference on
Conference_Location :
San Diego, CA
Print_ISBN :
978-1-4244-1765-0
Electronic_ISBN :
1522-4880
DOI :
10.1109/ICIP.2008.4712328