참조영상의 색 조화 패턴을 활용한 색상 변환 기법 = Color Reproduction Techniques Based on Color Harmonized Pattern of Reference Image

다국어 초록 (Multilingual Abstract)

A color reproduction technique is a study in which the color information of the reference image is expressed in the input image. Previous studies makes natural color transferred results by allowing the input image to have a distribution of colors with reference images. However, if reference images are used with works by famous artists, their characteristics are not properly expressed. This is because works by famous artists use limited colors, and color harmonized pattern. In this study, we propose novel color reproduction techniques which use artist's work as a reference image and reflects the characteristics of the reference image. Our algorithm consists of two stages. The first step is the reference image analysis phase. For this purpose, the representative colors used in the reference images were extracted using K-means clustering algorithms. And the color harmonization patterns that the work has were analyzed by analyzing what forms are composed between the representative colors of the reference image. The second stage is the color transfer phase using the analysis information. For this, we use image saliency algorithm to extract the most focusing color of the input image. Image saliency is a study that extracts the most visible areas of the image, making it the most visible color of the input images viewed by the area with the highest visibility. The rest of the region applied the color harmonization pattern of the reference image. Our algorithm can easily generate result images that express the color harminized pattern effects of works used as reference images.

참고문헌 (Reference)

1 D. Chen, "Visual Saliency Detection" 44 (44): 57-68, 2016

2 Y. Kim, "Saliency-guided enhancement for volume visualization" 12 (12): 925-932, 2006

3 L. Liang, "Real-time texture synthesis by patch-based sampling" 20 : 127-150, 2001

4 T. Pouli, "Progressive histogram reshaping for creative color transfer and tone reproduction" 81-90, 2010

5 J. P. Collomosse, "Painterly rendering using image salience" EGUK 122-128, 2002

6 C. H. Lee, "Mesh saliency" 24 (24): 659-666, 2005

7 Y. W. Tai, "Local color transfer via probabilistic segmentation by expectation maximization" 217-226, 2005

8 D. Lischinski, "Interactive local adjustment of tonal values" 646 (646): 646-653, 2006

9 A. A. Efros, "Image quilting for texture synthesis and transfer" 341-346, 2001

10 Q.C. Tian, "Histogram-based color transfer for image stitching" 3 (3): 38-, 2017

11 H. Jonathan, "Graph-based visual saliency" 19 : 545-552, 2007

12 X. Xiao, "Gradient-preserving color transfer" 28 : 1879-1886, 2009

13 M-M. Cheng, "Global contrast based salient region detection" 409-416, 2011

14 M. Ashikhmin, "Fast texture transfer" 23 (23): 38-43, 2003

15 C, Wen, "Example-based multiple local color transfer by strokes" 27 (27): 1765-1722, 2008

16 E. Reinhard, "Color transfer between images" 21 : 34-41, 2001

17 L. Neumann, "Color style transfer techniques using hue, lightness and saturation histogram matching" 111-122, 2005

18 D. Cohen-Or, "Color harmonization" 624-630, 2006