为了有效地滤除红外图像中的噪声，提出了一种小波域多方向自适应加权伪中值滤波算法。该算法首先对红外噪声图像的各高频分解子图像分别进行噪声点检测和标记；然后根据各子图像中像素点分布特征分别设计出4类具有多方向性的滤波模板进行自适应加权滤波；最后将低频分解子图像与滤波后的各小波高频分解子图像进行重构。分别将中值滤波（MF）、伪中值滤波（PMF）、极值中值滤波（EMF）、加权中值滤波（WMF）、以及本文算法应用于标准测试图像以及红外图像去噪，并引入峰值信噪比（PSNR）、平均绝对误差（MAE）进行去噪效果评定。标准测试图像和红外图像仿真结果表明，该算法性能明显优于PMF，且相对于与其余几类同类型算法而言，也具有一定的优势。%In order to filter the noise in infrared image, a multi-direction adaptive weighted pseudo median filtering algorithm is proposed based on wavelet transform. Firstly, the salt & pepper noise image is conducted by wavelet transform, and the high-frequency sub-images and low-frequency sub-image are obtained. The noise distribution areas of the high-frequency sub-images are detected and labeled effectively. Then, according to the characteristics of the ground objects and the features of the directionality of the high frequency wavelet decomposition sub-images, four kinds of directional filtering templates are respectively designed so as to deal with the noise through adaptive weighted filtering. Finally, low-frequency sub-image and high-frequency sub-images are reconstructed. The median filtering(MF), pseudo median filtering(PMF), extreme median filtering(EMF), weighted median filtering(WMF) and the algorithm in this paper are used to filter the salt & pepper noise in standard test image and infrared image. Peak signal to noise ratio(PSNR) and mean absolute error (MAE)are adopted to evaluate the performance of above filtering algorithms. The experimental results show that the performance of the algorithm in this paper is better than the others.
展开▼
