资源描述：
《双边滤波毕业论文》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。
1、摘耍双边滤波方法(Bilateralfiltering)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
2、素的平均灰度值去代替要处理图片的毎个像素点的值，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。屮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乂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。本文主要介绍儿种常见的滤波方法，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AbstractBi
3、lateralfilteringmethod(Bilateralfiltering)ismadebasedonGsussfilter,mainlyforGaussGaussfilterweightswillbedirectlywiththeconvolutionofimageinformationforimagefilteringprinciple,theoptimalfilterweightsandtheimagebrightnessintotheGaussfunctioninformationproduct,theopt
4、imizedweightcoefficienttoconvolutionwiththeimageinformation,sothatwillbeabletofiltertakingintoaccounttheimageinformationintheimageedgeinformation,sothatthenormalGaussfilteredimageisblurrededgetomaintainaclearandsmootheredges.Thismethodforcolorandgrayscaleimagesofth
5、efilterareapplicable,withstrongapplicability.Thetraditionalfilteringtechnologiesmeanfiltering,medianfiltering,averagefilteringistousefiltermasktodeterminetheneighborhoodoftheaveragegrayvalueofpixelsinsteadofdealingwithpicturestoeachpixelvalue,itcouldsignificantlyno
6、isereduction,However,theedgeoffuzzymeanfiltercancausenegativeeffects.Themedianfilteristhepixelgrayscalemedianneighborhoodinsteadofthepixelvalue,whichdealswithimagesisbetterthanthemeanfilter,itismoresuitableforPittaveragefiltertoremovesaltandpeppernoise.Bilateralfil
7、teringisanadditivecanreducetheimagenoisecanalsoclearedgefilteringtechnology,itcanalsousethespatialneighborhoodofpixelinformationproximityandbrightnesssimilarityinformationtoprocessimages,isarelativelygoodthefilter.Thispaperdescribesseveralcommonfilteringmethods,hig
8、hlightsomeofthebilateralfilterandmedianfilterwithhimcomparedtotraditionalmethods.Keywords:Bilateralfiltering,Gaussiannoise,Edgepreserving,Imagede
