¡¡Chinese Journal of Computers   Full Text
  TitleAn Automatic Generation Method of Multi-Styles Portraits Based on the And-Or Graph Representation
  AuthorsMIN Feng SANG Nong
  Address(Institute for Pattern Recognition and Artificial Intelligence, Huazhong University of Science and Technology, Wuhan 430074)
  Year2009
  IssueNo.8(1595¡ª1602)
  Abstract &
  Background 		Abstract This paper presents an automatic generation method of multi-styles portraits based on the And-Or graph representation. The And-Or graph is a generative model, which separates the structure and style of portraits and accounts for the variability of portraits. Adopting the generative model, the method can automatically generate a set of portraits in different styles from a frontal face image. The method decomposes portrait into different components. Each component has a number of distinct sub-templates as leaf-nodes in the And-Or graph for portrait. The And-Or graph for portrait is like a ¡°mother template¡± which produces a large set of valid portrait configurations¡ª¡ª¡°composite templates¡± that are made of a set of sub-templates. The method benefits from a number of template dictionaries for portrait components in different styles. Therefore, it is convenient to change the styles of portrait by changing the template dictionary. Experimental results demonstrate the effectiveness of the method.
Keywords And-Or graph; template; automatic portraiture; non-photorealistic rendering
Background This paper is supported by the National Natural Science Foundation of China under project ¡°The Study of Automatic Human Portrait and Cartoon Animation Generation¡± with Grant No.60672162, The National High Technology Research and Development Program (863 Program) (Grant No.2007AA01Z166), Program of New Century Excellent Talents in University (NCET-05-0641). The project focuses on face representation and computation, the technology of automatic portrait and cartoon animation generation from a human image or video. This work is aimed at a number of applications, such as low bit portrait communication in wireless platforms, cartoon sketch and canvas in non-photorealistic rendering, portrait editing and make-up. This work is done at the Lotus Hill Research Institute and Institute for Pattern Recognition and Artificial Intelligence, Huazhong University of Science and Technology. The author thanks Zhu Song-Chun for guidance.