The Study Of Style Transformation For Line Drawings

Since the 1990s, Non-photorealistic rendering has become one of the hottest research topics in the field of graphics rendering. It focuses on how to present artistic characteristics and simulate artistic effect of the artwork with some certain styles, which is different from the Photorealistic Rendering that aims to simulate the real world.Line Drawing is an artistic form which has simple ways of creation and sufficient expressiveness. It is popular and also plays an important role in the social production and living. The extraction and transformation of the styles of line drawings has become an important part and one of the hottest research topics in the field of Non-photorealistic rendering. This paper focuses on how to extract the style of lines from the sample drawing and then apply the style to target drawing that doesn't have this style. Finally, the style of the drawing will be similar to the style of the example drawing. In order to attain this goal, based on analyzing and researching the characteristics, a style translation method of the line drawings is put forward and a prototype system based on the framework is implementedThe framework consists of parts of capturing styles and styles synthesis. The methods to achieve generally as follow, firstly extracts every line from the example line drawing by a series digital image processing methods. Then styles of lines will be qualified according to the model established before. Finally, apply the style to another line drawing which has got rid of its style.The result of experiments showed that this framework displayed good results in the style transformation and has the following merits. Firstly, this method can overcome the shortcoming of multiresolution curve which only can deal with geometry property of lines. Secondly, it can largely lower the demands for number of example lines. Thirdly, it will be able to set several exaggeration degrees of vision properties, thus the end of adjustable style can be achieved. Actually, there are still some deficiencies. With the improving of algorithm implementation in the future work, these problems will be solved eventually.