摘要
图像抽象化绘制旨在以模糊、抽象的艺术手法展现图像。提出了基于边缘切向流与颜色量化的图像抽象化绘制方法,首先,在RGB(Red,green,blue)颜色空间提取各颜色通道的梯度,并对其进行融合,对融合后的梯度图像求边缘切向流,并使用高斯差分滤波(flow-based difference-of-gaussians,FDOG)方法提取图像的特征线条;然后,对原始图像用非线性方法进行平滑处理,并在(hue-saturation-intensity,HSI)颜色空间进行量化;最后,将量化图像和特征线条图像进行融合,得到原图像的抽象化图像。实验证明该方法增强了图像轮廓的主要特征并忽略了局部细节,突出了图像的层次感,使图像简洁清晰,抽象化效果显著。
The aim of image abstraction is to show the image through means of blurring and abstract artistic expression.A method of image abstraction based on edge tangential flow and color quantization is proposed.Firstly,in the RGB color space,the gradient of each color channel is extracted and fused,abstracting the Edge tangential flow of the fused gradient image,and the feature lines of the image are extracted by using the FDOG filtering method.Then,the original image is smoothed by the nonlinear method,and in the HSI color space the image is quantized.Finally,the quantized image and the feature lines image are fused to obtain the image of the abstract image.Experiments show that the proposed method enhances the main features of the image contour and ignores the local details,highlighting the image level,so that the image is concise and clear,the abstract results are significant.
作者
陈烽
薛茹
CHEN Feng;XUE Ru(College of Information Engineering,Xizang Minzu University,Xianyang,Shanxi 712082,China)
出处
《南华大学学报（自然科学版）》
2019年第5期78-83,共6页
Journal of University of South China：Science and Technology
基金
西藏自治区自然基金项目(2015ZR-13-24)
西藏民族大学青年学人培育计划资助项目(16MYQP05)
关键词
边缘切向流
颜色量化
FDOG
图像抽象化
edge tangent flow(ETF)
color quantization
FDOG
image abstraction
