A FUNCTION APPROXIMATION METHOD FOR IMAGES WITH GRADING REGIONS
KENTARO MIYAMOTO, Tetsuo Kamina, Tetsuo Sugiyama, Keisuke Kameyama, Kazuo Toraichi, Yasuhiro Ohmiya · International Journal of Image and Graphics · 2009
In the last decade, many function-approximation methods for illustration images have been proposed. However, these methods cannot process more complex full color images effectively, even when some simple regularity of color transitions such as gradations appear in images. In this paper, we propose a rigorous function-approximation method that is applicable for gradation images. This method performs image segmentation that recognizes a gradation pattern as one region. This segmentation is performed by recursively approximating the tones of colors using multiple regression analysis of 2-variable functions. We compare the proposed method and the conventional methods with regard to the quality of enlarged images and the efficiency of image description and reproduction. The experimental results show that our approach is higher in quality, smaller in data size, and faster in computational costs than the conventional methods.