An image segmentation method for function approximation of gradation images

  • Authors:

  • Kentaro Miyamoto;Tetsuo Kamina;Tetsuo Sugiyama;Keisuke Kameyama;Kazuo Toraichi

  • Affiliations:

  • Graduate School of Systems and Information Engineering, University of Tsukuba, Tsukuba, Ibaraki, Japan;Center for Tsukuba Advanced Research Alliance, University of Tsukuba, Tsukuba, Ibaraki, Japan;Center for Tsukuba Advanced Research Alliance, University of Tsukuba, Tsukuba, Ibaraki, Japan;Center for Tsukuba Advanced Research Alliance, University of Tsukuba, Tsukuba, Ibaraki, Japan;Center for Tsukuba Advanced Research Alliance, University of Tsukuba, Tsukuba, Ibaraki, Japan

  • Venue:
  • SPPRA'06 Proceedings of the 24th IASTED international conference on Signal processing, pattern recognition, and applications
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

Function-approximated images are useful for quality-maintained affine-transform. However, it is difficult for conventional approximation methods to accurately function-approximate images including numerous small color regions such as gradations, because no appropriate segmentation is performed. We propose a new image segmentation method for function-approximation of gradation images and its description format. In this method, a gradation pattern in a image is recognized as a region by a new labeling method using multiple regression analysis of 2-variable functions. Pixel values in segmented color regions can be reproduced by using the contour and region approximation. The experiments show that in this method we can reduce the processing time and the file size becomes compact. By evaluating approximation accuracy by PSNR, it is proved that our approach improves the drawing accuracy.