AUTOMATIC CARTOON COLORING SYSTEM
Mohammad Gamal Eldeen Ahmad Mandour;
Abstract
The importance of animated images lies upon its participation in fields like education, entertainment, and advertisement. The traditional method in developing an animated scene is very time consuming and requires a lot of effort. It depends on manual drawing and coloring of individual frames that form the scene.
This thesis proposes an easy and automated technique for coloring the animated scenes. This proposed technique saves a lot of time and effort required by the traditional method. It is based on tracking objects in a cartoon scene and providing the cartoonist with a fast and simple tool for applying a proposed color scheme to those tracked objects. The propose technique will only require manual coloring of key frame to acquire the coloring scheme. The technique will automatically color the whole scene afterwards. The correspondence between objects in the scene is established by tracking some simple features such as the boundaries, area size, and center of gravity. Other advanced features like central moments are also included in a goal to enhance the coloring process.
Correspondence among object details (defined by closed contours) in different frames is made in two steps; first the closest candidates are identified by minimizing the Euclidean Distance between objects in the features vector space. Second, the most appropriate among those candidates are selected. Two algorithms are studied for finding correspondence, the first one is Link Break and Relink (LBR). This algorithm is based on breaking multiple links leaving the minimum one to generate a solution. The second algorithm, which is based on the Genetic Algorithm is used as an optimization tool for selecting an assumed better candidates.
The proposed technique is evaluated on data samples of 4304 objects in 113 frames that compose 10 scenes. The new technique fulftlls more than 95% percentage success in coloring the test data. This percentage reflects a minimum rework needed to rectify miss-assigned colors.
This thesis proposes an easy and automated technique for coloring the animated scenes. This proposed technique saves a lot of time and effort required by the traditional method. It is based on tracking objects in a cartoon scene and providing the cartoonist with a fast and simple tool for applying a proposed color scheme to those tracked objects. The propose technique will only require manual coloring of key frame to acquire the coloring scheme. The technique will automatically color the whole scene afterwards. The correspondence between objects in the scene is established by tracking some simple features such as the boundaries, area size, and center of gravity. Other advanced features like central moments are also included in a goal to enhance the coloring process.
Correspondence among object details (defined by closed contours) in different frames is made in two steps; first the closest candidates are identified by minimizing the Euclidean Distance between objects in the features vector space. Second, the most appropriate among those candidates are selected. Two algorithms are studied for finding correspondence, the first one is Link Break and Relink (LBR). This algorithm is based on breaking multiple links leaving the minimum one to generate a solution. The second algorithm, which is based on the Genetic Algorithm is used as an optimization tool for selecting an assumed better candidates.
The proposed technique is evaluated on data samples of 4304 objects in 113 frames that compose 10 scenes. The new technique fulftlls more than 95% percentage success in coloring the test data. This percentage reflects a minimum rework needed to rectify miss-assigned colors.
Other data
| Title | AUTOMATIC CARTOON COLORING SYSTEM | Other Titles | نظام تلوين آلى للرسوم المتحركة | Authors | Mohammad Gamal Eldeen Ahmad Mandour | Issue Date | 2001 |
Recommend this item
Similar Items from Core Recommender Database
Items in Ain Shams Scholar are protected by copyright, with all rights reserved, unless otherwise indicated.