基于图像空间的素描效果生成技术

Research of generating sketching style from original image

目前大部分非真实感绘制技术都利用各种不同大小及形状的笔画模拟人工绘画的过程,实现起来比较复杂。提出了一种基于图像空间的无笔画的素描效果生成算法,只需将一幅数字图像作为输入,然后对其进行滤波处理,使得处理后的图像在去噪的同时可以消除光照所带来的影响,同时可以将特征像素进行分离;在此基础上再分别进行边缘检测和二值化;最后将边缘检测结果和二值图像相结合,从而得到相应的素描效果。实验结果表明,该算法可以快速有效地生成素描效果,且实现简单。

Most current Non-Photorealistic Rendering （NPR） techniques perform rendering using strokes with different size and shape. This is largely to mimic hand painting practice. A non-stroke-based method was proposed for generating sketching style of painting from original image. A feature preserved filter was used to remove the illumination effect and separate feature pixels from others. The result image was transformed into binary version and edges were detected from the filtered image. Finally, the target sketching style was obtained by removing noises from binary version according to the edge information. At last, experimental results demonstrate that the new method can quickly and efficiently generate sketch effects, and the implementation is simple.