一种基于轮廓提取与纹理分析的图像着色方法  

Texture-based image colorization using directional sketch

作　　者：金舟[1] 孙济洲[1] 张怡[1] 

机构地区：[1]天津大学计算机科学与技术学院,天津300072

出　　处：《计算机应用研究》2011年第12期4757-4759,4762,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(61003201);天津市科技支撑计划重点资助项目(10ZCKFGX00800)

摘　　要：从分析卡通、漫画图像的构图方法入手提出一种基于颜色扩展的图像着色方法,通过提取图像轮廓边界形成着色方向场指导上色过程,利用Gabor小波实现对图像中非相邻纹理区域的分析与匹配。实验结果说明该方法通过简洁的色彩关系和明暗层次突出了卡通、漫画图像适度简化和夸张的艺术特点,不相邻区域纹理匹配有效简化了细节部分添加色彩标记的用户交互复杂度。This paper proposed a novel scribble based method to colorize comic images. The colorization process was guided with the orientation field which was generated from the primal sketch. The pattern-based features were measured by using the statistical feature in Gabor wavelet domain. A variety of results indicate that the directional sketch model provides a straightforward way to present fine details in comic images, and brings significant time savings by propagating colors over various complex texture regions.

关 键 词：轮廓提取 纹理分析 渐变着色 笔迹保持 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]