基于多尺度笔刷的分层油画风格化

Layered Oil Painting Stylization with Multi-Scale Brushes

基于图像的油画风格化绘制是计算机图形学领域非真实感绘制研究的热点之一.为了进一步提高图像油画风格化的质量,提出了一种基于多尺度笔刷的分层图像油画风格化绘制算法.该算法模拟艺术家的油画绘制过程,采用不同尺度的笔刷按照从粗到细的顺序逐层绘制.在每层笔刷绘制中,首先使用增量Voronoi序列采样点和图像切线方向场确定笔刷流线,然后结合笔刷形状与笔刷高度场进行纹理贴图,得到最终的图像油画风格化绘制结果.通过与现有算法比较,文中算法不仅能模拟真实的油画绘制过程,而且生成的油画效果层次感更强,充分体现了图像的结构特征和油画细节.

Image based oil painting stylization is a popular research topic in the domain of non-photorealistic rendering.In this paper,we propose a novel layered oil painting stylization algorithm which applies multi-scale brushes to improve the quality of oil painting stylization.Considering the real painting procedure of artists,we adopt multi-scale brushes to obtain oil painting effects by successively combining layers where brushes of large scales to small ones are employed.In each layer,we use incremental Voronoi sequence and image tangent orientation field to create brush lines.Stylized oil painting brush strokes are synthesized by texture mapping according to brush shape and brush height field.Our method can simulate the painting procedure of artists.Compared with existing methods,better oil painting results are achieved where both structures from input images and details from synthesized brushes are exhibited.