摘要
提出了一种基于非真实感绘制的漫画艺术效果绘制算法。算法首先将输入图像从RGB色彩空间转换为LAB色彩空间,通过双边滤波及色彩量化的方法对输入图像进行处理,获得连续的、局部区域色彩一致的结果图像,可通过权值等参数对滤波及量化程度进行控制。其次,采用DOG滤波器和阈值处理求取输入图像的边缘信息,获得连续的、清晰的边缘。最后将边缘图像融合到量化结果图像中,获得最终的漫画艺术效果。从最终的漫画艺术绘制效果看出,该算法简单、易于实现。
An effective method for non-photorealistic rendering for computer generated images with cartoon artistic works from 2D images is provided in this paper. The proposed methods is inspired by bilateral filter and DOG filter. The color space of source image is transferred from RGB to LAB, and bilateral filter can be used to reduce contrast in low contrast regions, and increase contrast in high contrast regions. Further, the color can be abstracted by soft color quantization to create cartoon-like effects. In addition, DOG filter and threshold are applied to obtain continuous and clear edge image, and this edge information can be merged to the cartoon-like effects finally. Experimental results show that our proposed method is simple and effective.
出处
《图学学报》
CSCD
北大核心
2016年第2期218-223,共6页
Journal of Graphics
基金
国家自然科学基金项目(61462093
61163019
61063009)
教育部博士点基金新教师类项目(20125301120008)
云南省应用基础研究计划项目(2014FA021
2014FB113)
云南省教育厅重点项目(2015Z012)
关键词
非真实感绘制
漫画
双边滤波
量化
DOG滤波
non-photorealistic rendering
cartoon
bilateral filter
quantization
DOG filter