改进型滤波器组与PCA结合的图像降噪算法  

Image Denoising Method with Combination of Modified Quincunx Filter Banks and Principal Component Analysis

作　　者：王佳宁[1] 

机构地区：[1]陕西学前师范学院计算机与电子信息系,西安710100

出　　处：《计算机与数字工程》2015年第4期572-575,共4页Computer & Digital Engineering

基　　金：陕西省教育厅科学研究专项(编号:14JK1179);陕西学前师范学院科研项目(编号:2013KJ068)资助

摘　　要：在图像的除噪处理中,为了能更好地处理图像的边缘及纹理处的噪声,论文提出了利用非均衡五点梅花采样滤波器组(nuQFB)结合主成分分析(PCA)的图像除噪方法。NuQFB克服了传统小波分解方向分辨率低和传统多方向滤波器组无多尺度分析的缺点,实现了对含噪图像的多尺度和多方向分解的统一,使分解后的图像具有十二个高频方向子带。并通过主成分分析的统计学特征提取方法对分解后的系数进行噪声能量估计完成图像的除噪处理。试验结果表明,与小波主成分分析的除噪方法相比,经该文方法除噪后的图像具有较好的视觉处理效果与峰值信噪比(PSNR)。In order to obtain better effect on image edge and texture for image denoising process, a method based on modified nonuniform Quincunx Filter Banks（nuQFB） with Principal Componet Analysis（PCA） is proposed. NuQFB contains twelve highpass subbands, and overcomes the drawbacks that traditional 2D-wavelet transform lacks muhidirectional infor- mation and directional filter banks lack multiresolution property. Which the combination of multiresolution and directional a- nalysis, Principal Componet Analysis method is adopted to estimate noise energy for image denoising. The experiment results show that the performance of the proposed method is obviously superior in both vision and Pulse Signal to Noise Ratio （PSNR）.

关 键 词：非均衡方向滤波器组 五点梅花采样 主成分分析 图像除噪 

分 类 号：TN911.73[电子电信—通信与信息系统]