绘制内容指导的铅笔画风格实现

Content Guided Pencil Drawing Style

为了让铅笔画纹理的色调更接近画家的绘制效果,提出一种内容指导的铅笔画色调生成算法,以改进铅笔画纹理的生成效果.首先对大量绘制内容不同的铅笔画分类后进行色调直方图统计,拟合出适合每种类别的直方图曲线,并对输入图像用与之同类别的直方图曲线进行色调匹配,得到对应的铅笔画色调图;然后用与输入图像大小相同的随机噪声图为基底,对噪声图的每个像素进行不同次数的迭代相乘后产生与色调图匹配且分布一致的噪声场;最后对输入图像进行区域分割,在与分割区域对应的噪声场中用LIC算法生成卷积长度及卷积方向不同的铅笔纹理图,实现分区域不同方向的铅笔纹理的绘制.实验结果表明,由于采用内容指导的分类绘制,该算法生成的铅笔画在色调上更加贴近真实画家绘制的铅笔画,且纹理形式多样,整体效果较好.

Many artistic styles of painting,such as pencil drawing,ink drawing,paint drawing,pen drawingetc.,can be generated by using non-photorealistic rendering techniques.This paper chooses pencil drawingas the main study object to carry out innovative rendering algorithm.Pencil drawing generation can be dividedinto two parts,line drawing and pencil texture.This paper is focused on how to improve the method ofgenerating a tone map used for pencil texture.First,a lot of pencil drawings painted by artists are collectedand classified into different categories,such as landscape drawings,architectural drawings respectively;thepencil histogram statistics to different categories are made,and the curve is used to fit the histogram.Thehistogram curve of input image is matched with the curve of same category to achieve the purpose of classificationdrawing.Second,the random noise map having the same size with the input image is multiplied bythe number of iterations tone map.The optimization algorithm is used to calculate the number of multiplication.Third,the images are segmented into different areas,LIC is used to form the texture map with differentlengths and directions which is also fitted for current area.The experiments show that the pencil drawinggenerated in this paper is more similar to the pencil drawing painted by artist.