摘要
针对中值滤波导致部分图像细节损失和均值滤波出现模糊现象,设计了一种适用于椒盐和高斯混合噪声的自适应滤波算法。该算法先用最小邻域的均值和阈值判断噪声类型,然后使用加权中值滤波处理椒盐噪声,再利用拉普拉斯算子和相应阈值判断图像边缘细节,最后对高斯噪声进行加权均值滤波。实验仿真结果表明,从图像视觉效果来看,相比单独使用中值和均值滤波降噪,自适应滤波算法对图像的还原效果更好,图像细节保存较好,模糊程度相对较弱,图像更清晰。通过对比峰值信噪比(PSNR)和均方误差(MSE),对混合噪声进行处理时,滤波算法的PSNR和MSE值优于中值和均值滤波,有效还原了噪声图像。整个算法是在最小邻域空间进行,易于实现,对混合噪声的处理效果较好,为图像处理的系统集成化设计提供了技术支持。
For the median filter resulting in some loss of the image detail and the mean filter appearing the vagueness, in this paper, presented is an adaptive filter algorithm suitable for the mixture noise with the impulse and Gaussian. First of all, the algorithm uses the average value in the minimum neighborhood and the threshold to determine the type of noise. Secondly, the algorithm uses the weighted median filter to process the impulse noise, then it uses the laplace operator and corresponding threshold to analyze the edge of image. At last, the algorithm uses the weighted mean filter to process the Gaussian noise. The simulation result shows, for the visual effect of image, comparing to separate median and mean filtering noise, the proposed algorithm is better in the effect on the reduction filter. The image is better preservation of detail, and weaker in the degree of blurring weaker, and some clearer in the clarity. Comparing the peak signal to noise ratio （PSNR） and mean square error （MSE）, when processing the mixed noise, the PSNR and MSE value of the proposed algorithm are better than the median and mean filter, and this algorithm effectively reduces the noise of image. The whole algorithm is processed in the smallest neighborhood space, easy to implement, and shows better treatment of the mixed noise.
出处
《半导体光电》
CAS
北大核心
2016年第4期568-572,共5页
Semiconductor Optoelectronics
关键词
降噪滤波
混合噪声
图像处理
最小邻域空间
de-noise filtering
mixed noise
image processing
the smallest neighborhoodspace
