摘要
针对使用统计方法对图像进行分区域去噪时,区域划分的阈值存在人为选择和整个去噪过程耗时过多的问题,提出了一种无需人为选择阈值参数的快速分区域去噪算法。对污染图像进行噪声估计;用两个矩阵分别存储在后期计算过程中反复使用的平方值,进行无需人为选择阈值参数的区域划分;分别用不同的策略对不同区域去噪,将区域去噪结果合成恢复图像。实验结果表明,该算法具有较快的运算速度,处理的图像有较高的峰值信噪比。
To solve the problems that threshold is selected manually in process of segmenting and much time is spent when statis- tical method is used to remove noise, a fast denoising algorithms based on regions is presented. Firstly, noise intensity per pixel is estimated approximatively and two matrixes which store the square of intensities of contaminated image and approximate noise per pixel respectively are built. Secondly, the image is partitioned into two regions with the exact threshold. Finally, the intensi- ties of two regions are handled separately, and then the results constitute the final denoising image. The experimental results show that the algorithm has the capacity of fast denoising, and the denoising image it processes has higher peak signal to noise ra- tio （PSNR）.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第4期1341-1346,共6页
Computer Engineering and Design
基金
四川省应用基础研究计划基金项目(2011JY0060)
关键词
图像分割
区域
统计方法
边缘检测
高斯噪声
去噪
image segmentatiom regions
statistical method
edge detectiom Gaussian noise
denoising
