Title :
Competitive image colorization
Author :
Kawulok, Michal ; Smolka, Bogdan
Author_Institution :
Inst. of Comput. Sci., Silesian Univ. of Technol., Gliwice, Poland
Abstract :
This paper presents a new method for image colorization based on manually added scribbles. We determine color propagation paths in the image by minimizing the geodesic distance from the scribbles using Dijkstra algorithm. After that, chrominance blending is performed to colorize the image. Our contribution lies in proposing the competitive approach for selecting an appropriate type of the path cost. Moreover, we introduce a modified chrominance distance calculated along each path. Benefits of these improvements are explained and justified. The experiments confirmed that the proposed algorithm yields better results compared with other well-established methods.
Keywords :
differential geometry; image colour analysis; Dijkstra algorithm; chrominance blending; color propagation paths; competitive image colorization; geodesic distance; manually added scribbles; Gray-scale; Hair; Image color analysis; Image segmentation; Optimization; Pixel; Probabilistic logic; chrominance blending; distance transform; image colorization;
Conference_Titel :
Image Processing (ICIP), 2010 17th IEEE International Conference on
Conference_Location :
Hong Kong
Print_ISBN :
978-1-4244-7992-4
Electronic_ISBN :
1522-4880
DOI :
10.1109/ICIP.2010.5653544