• DocumentCode
    3592354
  • Title

    Higher Order Polynomial Demosaicing

  • Author

    Hernandez-Lopez, Jose-Juan ; Flores-Barranco, M. Magali ; Ibarra-Manzano, Mario Alberto ; Ledesma-Orozco, Sergio Eduardo ; Almanza-Ojeda, Dora Luz

  • fYear
    2014
  • Firstpage
    14
  • Lastpage
    19
  • Abstract
    Traditional digital cameras use a single charged coupled device (CCD) to capture images. In order to capture all color components (RGB), a light filter is located between the lens and the sensor, to separate the light and capture only one of these colors per pixel, this filter is known as Bayer filter. The resultant image contains the coded color components in two green pixels, one red pixel and one blue pixel per quartet, and a demosaicing process is needed to recover the coded color components. In this paper, we propose a novel methodology to perform this demosaicing on Bayer images, utilizing a higher order polynomial interpolation. We evaluate the second and third order polynomial, but, any order polynomial is easily implementable. Finally, we compare our method against well known methods such as simple interpolation demosiacing and nearest neighbor.
  • Keywords
    image capture; image coding; image colour analysis; image filtering; image segmentation; interpolation; Bayer filter; Bayer images; RGB color component; blue pixel; coded color components; digital cameras; green pixels; higher-order polynomial demosaicing; higher-order polynomial interpolation; image capture; image sensor; lens; light filter; red pixel; second-order polynomial; third-order polynomial; Image color analysis; Image edge detection; Interpolation; Mathematical model; Polynomials; Splines (mathematics); Bayer filter; Demosiaicing; Image processing; Nearest Neighbor; Polynomial interpolation;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Mechatronics, Electronics and Automotive Engineering (ICMEAE), 2014 International Conference on
  • Type

    conf

  • DOI
    10.1109/ICMEAE.2014.21
  • Filename
    7120839