Research On Image Colorization Algorithms

Image color editing technology as the hot topic among computer graphics and computer vision fields, are widely used for animation film, television, culture heritage, entertainment, and advertising. In this paper, we major research on two relative important algorithms among image color editing field, which are image colorization and recoloring. In the paper, we first review the background and the research status of the image colorization and recoloring, and then we introduce the image colorization method of manifold-preserving image colorization with nonlocal estimation. We optimize the color distribution with an energy minimization between a pixel and texture-similar neighbors in the feature space which proposed by this paper. Experimental results show, This method have a big improvement in the efficiency and effectiveness. Meanwhile, in order to avoid artifacts along high-contrast boundaries, we introduce structure-aware nonlocal optimization framework for image colorization. The method defines a new high feature space, and the feature space contains not only image coordinates and intensities but also statistical texture features obtained with the direction-aligned Gabor wavelet filter. In the high feature space, by implementing the nonlocal principle, colors can effectively propagate to not only local intensity-continuous region, but also remote texture-similar regions without explicit image segmentation. Structure maps are utilized to scale texture features to avoid artifacts along high-contrast boundaries. Experimental results show that this method also can be used in selective image recoloring and decoloring, and progressive color editing. Moreover, we proposed the image recoloring using geodesic distance based color harmonization. In this method, we proposed a new strategy of searching the best harmonic scheme to find the optimal template which well matches the hue distributions of the image. And we use the geodesic distance technology to reassign hues outside the harmonious sectors to inside the sectors to obtain the pleasing recolored image. Facts proved, those method proposed by this paper, can get the better results and more efficiency than other methods in the same condition, and have a high practical value.