Automated generation of graphic sketches by example

  • Authors:
  • Michelle X. Zhou;Min Chen

  • Affiliations:
  • IBM T. J. Watson Research Center, Hawthorne, NY;IBM T. J. Watson Research Center, Hawthorne, NY

  • Venue:
  • IJCAI'03 Proceedings of the 18th international joint conference on Artificial intelligence
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

Hand-crafting effective visual presentations is time-consuming and requires design skills. Here we present a case-based graphic sketch generation algorithm, which uses a database of existing graphic examples (cases) to automatically create a sketch of a presentation for a new user request. As the first case-based learning approach to graphics generation, our work offers three unique contributions. First, we augment a similarity metric with a set of adequacy evaluation criteria to retrieve a case that is most similar to the request and is also usable in sketch synthesis. To facilitate the retrieval of case fragments, we develop a systematic approach to case/request decomposition when a usable case cannot be found. Second, we improve case retrieval speed by organizing cases into hierarchical clusters based on their similarity distances and by using dynamically selected cluster representatives. Third, we develop a general case composition method to synthesize a new sketch from multiple retrieved cases. Furthermore, we have implemented our casebased sketch generation algorithm in a user-system cooperative graphics design system called IMPROVISE, which helps users to generate creative and tailored presentations.