Font Size: a A A

Research On Cartoon Rendering Technique Of NPR

Posted on:2011-12-17Degree:MasterType:Thesis
Country:ChinaCandidate:G Q HuFull Text:PDF
GTID:2178360308481137Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
The technology of Non-photorealistic rendering (NPR) develops quickly in recent years. Unlike traditional rendering method, it mainly simulates certain artistic styles, such as sketch, watercolor, oil painting, etc. This rendering technique completes a perfect combination of computer science and art, makes up the defect of artistic beauty in the realistic rendering, and brings people deeper appeal from the psychological and visual.The cartoon with bright color and unique style has become a popular visual culture today. However, there are various technical difficulties in hand-drawn cartoon. Thus, the technology that generates cartoon images and videos by computer simulation is more and more attention.This paper reviews the development process of non-photorealistic rendering techniques, and discusses mainly present some non-photorealistic rendering techniques and the methods of video stylization which is a significant branch of non-photorealistic rendering techniques. Then the paper analyzes some existed problems in methods of cartoon rendering, such as temporal and spatial discontinuity, real-time poor, modeling complex and requiring skilled user participation. Therefore, we propose an effective NPR technique, which improves the above problems, to auto-generate cartoon. At last,we try to apply our method to video frames.This thesis includes the following works:First, a bilateral filter is used to abstract image by modifying the contrast of visually important features, namely luminance and color opponency. Then, to simulate the cartoon coloring style, the technology called color quantization is used on abstracted image.Second, a non-photorealistic rendering technique that automatically extracts stylized edge map from 2D images, aiming to facilitate effective visual communication is presented. We introduce an anisotropic DoG (Difference-of-Gaussians) filter that can extract a set of coherent, smooth, and stylized lines from 2D images. These lines effectively capture and convey important shapes and details in the image.Third, the final cartoon rendering result is obtained by adding stylized edge map to the image color-quantized. To some extent, the result not only has the cartoon aesthetic but also has better visualization effect.Forth, the method based on 2D images is extended to process video frames and gets the satisfied result.
Keywords/Search Tags:cartoon rendering, image abstract, color quantization, stylized lines extraction
PDF Full Text Request
Related items