融合B样条小波变换的图像铅笔画生成改进方法

Improved method of image pencil drawing generation by fusing B-spline wavelet transform

彩色铅笔画自动生成算法是当前非真实感绘制技术研究热点之一。针对目前轮廓线条细节丢失的问题,该文提出融合B样条小波变换与二次卷积方法,得到的线条轮廓图保留更多细节信息且线条具有交叉感;线积分卷积(Line integral convolution,LIC)方法使用随机白噪声产生的纹理不清晰,且执行速度慢,该文在亮度图上添加高斯噪声替换随机二值白噪声,利用随机数控制噪声点,并应用运动模糊产生清晰纹理。实验证明,该算法得到的铅笔画效果与现存铅笔画生成算法相比,保留了更多的细节,更加接近实际艺术家作品风格。

The automatic generation algorithm of color pencil drawing is one of the research focuses in the field of non-photorealistic rendering(NPR).The details of the edge lines are often missing and the edge lines are not clearly shown.To solve the problems,the combination of B-spline wavelet transform with re-convolution is proposed.Strokes retain more details and show broken strokes with cross-like effect at junctions.The line integral convolution(LIC)method performs too slow and generates unclear texture from random white noise.Gaussian noise is added to the brightness map to replace the random binary white noise,random numbers are used to control noise points,and the motion filter can generate clear texture.The experimental results show that:compared with the other algorithms,the proposed algorithm preserves more details and approximates the actual artist’s work style.