1、 本科毕业论文(设计)本科毕业论文(设计) (2014 届届) 基于双边滤波的图像去噪方法 院 系 电子信息工程学院 专 业 通息工程 姓 名 指导教师 讲师 2014 年 4 月 学号:1008431110 I 摘 要 双边滤波是非线性的滤波方法,是结合图像的像素值相似度空间邻近度和空间领近度 的一种折衷处理,同时考虑灰度相似性和空域信息,达到保边去噪的目的。双边滤波具有 简单、非迭代、局部的特点。双边滤波器的好处是可以做边缘保存,一般过去用的维纳滤 波或者高斯滤波去降噪,都会较明显地模糊边缘,对于高频细节的保护效果并不明显。双 边滤波比高斯滤波多了一个高斯方差,它是基于空间分布的高斯滤波函
2、数,所以在边缘附 近,离的较远的像素不会影响到边缘上的像素值,这样就保证了边缘附近像素值的保存。 但是由于保存了过多的高频信息对于彩色图像里的高频噪声,双边滤波器不能够彻底的滤 掉,只能够对于低频信息进行较好的滤波。其具体的操作方法有两个,第一个是高斯模版, 用个模板对图像中的每一个像素值进行扫描,然后把某一点和其邻域内像素的加权平均值 代替那一个中心的值高斯滤波器是根据高斯函数的形状来选择其权值的线性平滑滤波器, 高斯滤波是线性平滑滤波的一种,最适合去除的噪声类型是服从正态分布的噪声。第二个 是以灰度级的差值作为函数系数生成的模板。然后这两个模板点乘就得到了最终的双边滤 波模板,最后得到双边
3、滤波处理后的图像。 关键词关键词:图像;去噪;双边滤波;高斯滤波 II Abstract The bilateral filter is a nonlinear filtering method, is the combination of image pixel value similarity space proximity and space brought a compromise approach degree, considering the gray similarity and spatial information, to achieve the purpose of edg
4、e preserving denoising. The bilateral filter has the advantages of simple, non iterative, local. The bilateral filter is good to do edge preservation,generally used Wiener filtering or Gauss filter to denoise, will obviously fuzzy edge, for the protection of high frequency detail is not obvious. Bil
5、ateral filtering than Gauss filter has a Gauss variance, it is Gauss filter function based on the spatial distribution, so near the edge, the pixel will not affect the farther to the pixel on the edge of the value, thus ensuring the preservation of edge pixel values. But because of the high frequenc
6、y information saved too much for the high frequency noise in the color image, the bilateral filter can not be completely filtered out, can only be better filtering for the low frequency information. The specific operation method has two, the first is Gauss template, scanning for each pixel in the image with a template, and then the weighted one point and its neighborhood pixels instead of the average value of a central value Gauss filters are linear smoothing filter