Title :
Painting style transfer and 3D interaction by model-based image synthesis
Author_Institution :
Inst. of Commun. & Inf. Technol., Zhejiang Gongshang Univ., Hangzhou
Abstract :
It presents a model-based image synthesis framework to accomplish goal to immerse a 3D object into a 2D model painting. By image synthesis and color transfer technique, the style of the model painting is transferred to the 3D image. The color transfer is automatically accomplished by basic color categories and statistical analysis of model painting and projected image. It proposed a new algorithm for relief image mapping. This algorithm greatly reduces search area when finding the intersection between rays and height-field by exploiting the min-max information. In the preprocessing step, we create min-max image to store min and max-min depth values that intersected by the rays passing through the texel. Thus it can reduce the search steps and then improve the performance and the visual quality. As the experimental results show the style of the model painting well and immerse the 3D object into a painting seamlessly.
Keywords :
image colour analysis; minimax techniques; rendering (computer graphics); statistical analysis; 3D interaction; color transfer technique; max-min depth values; min-max image; min-max information; model-based image synthesis; painting style transfer; relief image mapping; statistical analysis; Augmented reality; Cybernetics; Feedback; Filters; Image color analysis; Image generation; Machine learning; Painting; Process design; Rendering (computer graphics); 3D object; Image synthesis; color transfer; painting style;
Conference_Titel :
Machine Learning and Cybernetics, 2008 International Conference on
Conference_Location :
Kunming
Print_ISBN :
978-1-4244-2095-7
Electronic_ISBN :
978-1-4244-2096-4
DOI :
10.1109/ICMLC.2008.4620917