• DocumentCode
    678818
  • Title

    Proportional Image Enlargement Using Combinations of Scaling and Carving Method

  • Author

    Somawirata, I. Komang ; Uchimura, Keiichi ; Koutaki, Gou

  • Author_Institution
    Comput. Sci. & Electr. Eng., Kumamoto Univ., Kumamoto, Japan
  • fYear
    2013
  • fDate
    2-5 Dec. 2013
  • Firstpage
    106
  • Lastpage
    110
  • Abstract
    This paper proposes the proportional image enlargement using hybrid methods. Hybrid method is combinations of scaling and carving methods. This method consists of two steps. The first step enlarges the source image to the same size with minimum size for height or width from the target image. In this step, we use a kernel scaling method which is resulted in proportional content image size. The second step is the full size image enlargement in the width direction. The important content in the image is maintained. The image energy is used to detect the significant part in the image. We slice the image by following the minimum energy from top to bottom. The interpolation pixel is implemented among slices of the image. We use rank-ordered mean filters to reduce jagged image, especially in the interpolation pixels. The experiments show the proportional image enlargement, and the aspect ratio of an image is changed.
  • Keywords
    image processing; interpolation; carving method; hybrid method; image content; image energy; image slicing; interpolation pixel; jagged image reduction; kernel scaling method; minimum energy; proportional content image size; proportional image enlargement; rank-ordered mean filter; source image enlargement; Computational efficiency; Dolphins; Equations; Image resolution; Interpolation; Kernel; Read only memory; carving; image enlargement; scaling; window kernel;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Signal-Image Technology & Internet-Based Systems (SITIS), 2013 International Conference on
  • Conference_Location
    Kyoto
  • Type

    conf

  • DOI
    10.1109/SITIS.2013.28
  • Filename
    6727177