발간년도 : [2016]
| 논문정보 |
|
| 논문명(한글) |
[Vol.11, No.6] Video based Cartoon Rendering using Back-map and Edge-map |
|
| 논문투고자 |
Jae-Khun Chang, Hyeon-Jin Park, Seung-Taek Ryoo |
|
| 논문내용 |
The purpose of cartoon style rendering is to produce traditional cartoon stylized look such as comic books and cartoon movies. The essential of cartoon rendering is the black outline (inking) and the solid color of the interior regions (painting). In this paper, we suggest video based cartoon rendering using temporal frame difference. Video based cartoon rendering has problems occurred by temporal coherence. One of these problems is difficult to find accurate ink-lines from moving objects. Also, waving effect in the boundary area of painting is appeared by illumination variation. To resolve waving effect, we use back-map applied the difference between the current frame and the reference frame in painting step. Corrected frame can be generated from blending reference frame and current frame using back-map. Corrected frame is transformed into color based segmented frame using the mean shift algorithm. To improve inaccurate boundary edge, we use edge-map obtained by adding current edge extracted frame to edge difference between previous and current edge extracted frames in inking step. Blurred frame was generated by bilateral algorithm from color segmented frame. To extract boundary edge, we use sobel edge detection algorithm. The result image can be produced by subtracting edge-map from color segmented frame. |
|
| 첨부논문 |
|
|
|
|
|
자료실 > 논문지 
