Color Demosaicking via Directional Weighted TV Sparse Representation on Spectral Domain

2017 
This paper works on the improving the performance of the color demosaicking method for the CFA (color filter array) cameras via directional weighted TV sparse representation. In this paper, the authors clearly model the spectral camera capture process influenced by the material and the geometry of the object surface, and suppose that the normalized color bands can be modeled by 2D piecewise constant functions. Furthermore, in order to describe the change of surface materials and discontinuities in surface geometry, we model the Laplacians of the image's norm matrix as a 2D piecewise linear function. Then we proposed an algorithm based on the directional weighted TV sparse representation on spectral domain, and extensive experiments have been done and show the availability of the strong prior in color demosaicking.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    15
    References
    0
    Citations
    NaN
    KQI
    []