文档介绍：摘要
双边滤波方法(Bilateral filtering)是基于Gsuss滤波方法提出的,主要是针对Gauss滤波中将Gauss权系数直接与图像信息作卷积运算进行图像滤波的原理,将滤波权系数优化成Gauss函数和图像亮度信息的乘积,优化后的权系数再与图像信息作卷积运算,这样就能在滤波的同时考虑到图像信息中的图像边缘信息,使图像在正常Gauss滤波后很模糊的边缘信息得以保持清晰,并且图像边缘更加平滑。此方法对于彩色和灰度图像的滤波均适用,具有很强的适用性。
传统的滤波技术有均值滤波,中值滤波等,均值滤波是利用滤波掩膜确定的邻域内像素的平均灰度值去代替要处理图片的每个像素点的值,这样处理可以明显减噪,然而均值滤波会导致边缘模糊的负面效应。中值滤波是将像素邻域内灰度的中值代替该像素的值,它处理图像的效果要优于均值滤波,它碧均值滤波更适合去除椒盐噪声。双边滤波是一种既可有效降低图像加性噪声又可以保持图像边缘细节的滤波技术,它能同时利用邻域内像素点的空间邻近度信息和亮度相似度信息去处理图像,是一种比较不错的滤波方法。
本文主要介绍几种常见的滤波方法,重点讲双边滤波方法,并将他同中值滤波等传统方法作比较。
关键词双边滤波高斯噪声边缘保持图像去噪
Abstract
Bilateral filtering method (Bilateral filtering) is made based on Gsuss filter, mainly for Gauss Gauss filter weights will be directly with the convolution of image information for image filtering principle, the optimal filter weights and the image brightness into the Gauss function information product, the optimized weight coefficient to convolution with the image information, so that will be able to filter taking into account the image information in the image edge information, so that the normal Gauss filtered image is blurred edge to maintain a clear and smoother edges. This method for color and grayscale images of the filter are applicable, with strong applicability.
The traditional filtering technologies mean filtering, median filtering, average filtering is to use filter mask to determine the neighborhood of the average gray value of pixels instead of dealing with pictures to each pixel value, it could significantly noise reduction, However, the edge of fuzzy mean filter can cause negative effects. The median filter is the pixel gray scale median neighborhood instead of the pixel value,whichdeals with images is better than the mean filter,it is more suitable for Pitt average filter to remove salt and pepper filtering is an additive can reduce the image noise can also clear edge filtering technology, it can also use the spatial neighborhood of pixel information proximity and brightness similarity information to