Research And Implement Of Quick Query System Based On Full-text Query

As a counterpart of Realistic Rendering，Non-Photorealistic Rendering(NPR)ismore focused on the simulation and reproduction of artistic effect of the objective world.NPR has undergone a great development as an independent branch of computergraphics since1990s, and the designing scope are no longer just aboutcomputer-aided art drawing, but also cover film, education, games, entertainment,animation and other industries.In the past, mostly researches on Non-Photorealistic Rendering were about brushmode constructing. After constructed the brush model, computer drew brushwork bythe given framework and direction. This technology can achieve good result, but itneeds more interaction and professional knowledge of art domain. It also requites theuser to adjust parameters frequently to achieve good result.Hertzmann described a new framework for processing images by example, calledimage analogies. The framework involves two stages: a design phase andapplication phase. By this framework, we can learn the filtered from examples, thenwe can apply it to a new target image to get an analogous art-style filtered result. Thisframework builds upon a great deal of previous work in several disparate areas,including machine learning, texture synthesis and non-photorealistic rending.In this paper, we research on several relate work of disparate area, includingnon-photorealistic rending, texture synthesis and machine learning. Then we willdescribe the algorithm which the Hertzmann framework used, and analyzed thedrawback it has. To improve the synthesized result, we adopt fast nearest neighborssearching algorithm based on lower bound tree. The result and synthesized time willbe shown through the experiment based on the system we build.