摘要
提出一种基于非局部均值(NL-Means)的自适应采样方法,在使用非局部均值对图像进行去噪时,使用双缓冲区来存储图像,消除滤波系数和噪声之间的相关性。利用两个缓冲区之间的差值构建出一个权值地图,作为自适应采样的依据,从而使得合成图像时能够有针对性的进行采样,最终优化直接光照的绘制效果,提高其绘制效率。实验结果表明,该方法优于香农熵等经典方法。
In this paper,an approach based on NL-Means is proposed for adaptive sampling. In order to eliminate the correlation between the filter coefficient and noise,it uses two buffers to store the images while image denoising. It can construct a weight map by the difference value between two buffers as a guild for adaptive sampling,thus the sampling can be more pertinence while rendering images. As a result the rendering efficiency and effect of direct illumination can be improved. The experimental results show that the method can perform better than classic ones.
出处
《信息技术》
2016年第5期112-115,共4页
Information Technology
