一种针对椒盐噪声的滤波算法  

An Improved Adaptive Filter for Removing Salt and Peppers Noise

作　　者：黄琪[1] 李鼎权 施涛[1] 刘会刚[1] 

机构地区：[1]南开大学电子信息与光学工程学院,天津300071

出　　处：《南开大学学报（自然科学版）》2015年第4期26-31,共6页Acta Scientiarum Naturalium Universitatis Nankaiensis

基　　金：国家自然科学基金青年科学基金(61401237);天津市自然科学基金(13JCQNJC01200);高等学校博士学科点专项科研基金(20130031120034);国家级大学生创新创业训练计划(201310055040)

摘　　要：为了消除图像在传输过程中所引入的椒盐噪声,采用了一种2级检噪的自适应滤波算法.该算法对经典自适应滤波算法在检噪和滤噪功能上进行了改进.首先,针对不同类型图像采取了不同类型的检噪方案,然后区分出噪声点和信号点,最后,该算法根据噪声密度的不同采取不同的滤噪方案而保留信号点.仿真实验结果表明,在不同噪声密度情况下,改进后的算法不仅更有效地滤除椒盐噪声,而且能很好地保护原图的边缘与细节.不仅如此,即使针对存在大量极值信号点的图像,提出的算法仍有较好的滤波效果:针对加入黑色方块的Lena图像,当噪声密度为0.7时,该算法滤波的PSNR达28.16dB,滤波时间仅为6.46s.因此,对于不同类型的图像,该算法均满足实时性和高效性2个方向的要求.In order to remove salt and pepper noise image in the transmission process, an adaptive noise filtering algorithm is proposed, which uses two steps to detect the noise pixels. The proposed filter algorithm is based on the adaptive filtering algorithm. But it changes a lot on noise detection and noise filter. First, the algo- rithm takes different types of noise detection scheme to different types of images. Then the algorithm distin- guishes noise pixels and signal pixels. Finally, the algorithm takes different noise filtering scheme according to different noise density while retaining the signal points. The experiment shows that in different noise density, the improved algorithm is not only more effective filtering salt and pepper noise, and can well protect the edges of the original image with the details. What is more, even if there are a large number of extreme signal pixels in the image, the proposed algorithm can still have a good filtering effect. For ＇Lena＇ image with the black box, when the noise density 0.7, the PSNR of the filtering algorithm is 28.16 dB, and the filtering time is 6.46 s. Therefore, for different types of images, the proposed algorithm can effective and efficiently remove the noise.

关 键 词：2级检噪 自适应 椒盐噪声 极值 图像 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]