Traditional median filer and mean filter algorithms can eliminate the impulse noise and Gauss noise in an image respectively, but when the image contains these two types of noise at the same time,their filtering effect are not ideal. In order to suppress the impulse noise and Gauss noise mixed in the image,a median weighted filtering algorithm based on UK-flag shaped sub-windows is proposed, which uses the ideas of the multistage median filtering to divide 3×3 filtering window into UK-flag shaped sub-windows and then non-impulse noise pixels and their medians is calculated. The normalized method is employed to calculate the weights of these median pixels so that each median pixel in sub-window and its corresponding weight is weighted. The results have been acquired via filtering output of center pixel in the filtering window. Experimental results indicate that it has good filtering performance for mixed noise images and keeps the image edge details fare well and that the filtering effects are superior to the traditional median filtering algorithm,mean filtering algo-rithm and MLM algorithm,which displays certain practical value.%传统的中值滤波和均值滤波算法分别对脉冲噪声和高斯噪声有良好的抑制作用,但当图像中同时含有这两类噪声时,它们的滤波效果均不理想.为了抑制图像中混有的脉冲噪声和高斯噪声,提出了一种基于米字型子窗口中值加权的滤波算法.该算法借鉴多级中值滤波的思想将3×3滤波窗口划分米字型四个子窗口,通过统计各子窗口内非脉冲噪声像素点来计算这些像素点的中值,采用归一化方法计算这些中值点的权值,并将各子窗口的中值及对应的权值进行加权运算,运算结果作为中心点的滤波输出.实验结果表明,所提出的算法对混有脉冲噪声和高斯噪声的图像去噪能力较强,且较好地保持了图像的边缘等细节,滤波效果优于传统中值滤波算法、传统均值滤波算法和多级中值滤波MLM算法,且具有一定的应用价值.
展开▼
