【正文】 點(diǎn)（散斑）噪聲嚴(yán)重影響著分割后圖像的質(zhì)量，極易帶來(lái)過(guò)分割或者欠分割的現(xiàn)象，因此，通過(guò)濾波器，濾除斑點(diǎn)（散斑）噪聲對(duì)之后腫瘤圖像的分割起到舉足輕重的作用。 針對(duì)超聲 乳腺腫瘤圖像，通過(guò)利用計(jì)算機(jī)圖像處理技術(shù)，將腫瘤區(qū)域區(qū)分出來(lái)，從而為醫(yī)生的診斷提供便利同時(shí)提高醫(yī)生的診斷準(zhǔn)確度，是現(xiàn)在對(duì)于腫瘤診斷的一個(gè)快速發(fā)展領(lǐng)域。然而由于其成像機(jī)制的限制，圖像中存在大量的斑點(diǎn)（散斑）噪聲，在很大程度上干擾了病灶的識(shí)別。超聲成像系統(tǒng)由于其非侵犯性、便攜性，在臨床上得到越來(lái)越多的應(yīng)用。 Based on the partial differential equation anisotropic diffusion method。 Image segmentation technology。 (2) based on partial differential equation anisotropic diffusion method。 最后通過(guò)測(cè)試每一種濾波算法 處理后的圖像，應(yīng)用基于 FCM 算法的圖像分割，根據(jù)腫瘤區(qū)域輪廓分割的精確度的不同結(jié)果，綜合濾波結(jié)果，得出 基于偏微分方程的擴(kuò)散模型與雙邊濾波模型更有利于 FCM 圖像分割技術(shù)分割出精確的腫瘤區(qū)域。詳細(xì)介紹了以上三類(lèi)濾波器中的 10 種典型濾波算法（包括：均值濾波、中值濾波、 LEE 濾波、 Frost 濾波、Kuwahara 濾波、雙邊濾波、各向同性 TV 模型濾波、各向同性 PM模型濾波、各向異性 模型濾波、小波分析濾波），分析每種濾波算法的特征，并對(duì)這 10 種濾波算法依次進(jìn)行 MATLAB 實(shí)現(xiàn)。 本科畢業(yè)設(shè)計(jì)（論文） 題目： 針對(duì)超聲乳腺圖像的多種濾波器性能比較研究 學(xué) 院 電子與信息學(xué)院 專(zhuān) 業(yè) 信息工程 學(xué)生姓名 XXX 學(xué)生學(xué)號(hào) 指導(dǎo)教師 XXX 提交日期 年 月 日 I 摘 要 本文 簡(jiǎn)要介紹了計(jì)算機(jī)圖像處理技術(shù)在超聲乳腺腫瘤圖像應(yīng)用的現(xiàn)狀，對(duì)超聲腫瘤圖像進(jìn)行分割提取出腫瘤區(qū)域的圖像分割技術(shù)的背景，闡述了對(duì)超聲腫瘤圖像進(jìn)行濾波預(yù)處理實(shí)現(xiàn)消除圖像中的斑點(diǎn)（散斑）噪聲及保留邊緣特征的濾波技術(shù)的重要意義。 大致將濾波算法分為三類(lèi) (1)基于局域統(tǒng)計(jì)特性的空域?yàn)V波； (2)基于偏微分方程的各向異性擴(kuò)散方法； (3)基于多分辨率分析的斑點(diǎn)噪聲消除算法。 將已用 MATLAB 實(shí)現(xiàn)的濾波器對(duì)超聲乳腺腫瘤圖像進(jìn)行濾波， 從降噪、平滑、細(xì)節(jié)保留腫瘤區(qū)域界定等方面進(jìn)行分析，根據(jù)算法特征和實(shí)驗(yàn)數(shù)據(jù)， 分析得出基于偏微分方程的擴(kuò)散模型與雙邊濾波模型對(duì)超聲腫瘤圖像的斑點(diǎn)（散斑）噪聲有較好的抑制作用，同時(shí)增強(qiáng)了邊緣特征。 關(guān)鍵詞 ：超聲腫瘤圖像預(yù)處理； 圖像分割技術(shù)； 基于統(tǒng)計(jì)特性的空域?yàn)V波；基 于偏微分方程的各向異性擴(kuò)散方法；基于多分辨率分析的斑點(diǎn)噪聲消除算法 II Abstract This paper briefly introduces the puter image processing technology in the ultrasound mammary gland tumour of the present situation of image application, this paper introduces the tumor of the ultrasonic image segmentation extract tumor area image segmentation technology background, this paper expounds the ultrasonic image preprocessing achieve tumor filter to eliminate image of spots (speckle) noise and keep edge character of filter technology of significance. Roughly will filter algorithm is divided into three categories (1) based on local area statistical properties of the spatial filtering。 (3) based on the analysis of the resolution spots noise cancellation algorithm. Detailed introduces three kinds of more than 10 kinds of typical filters filter algorithm (including: average filtering and median filtering, LEE filtering, Frost filtering, Kuwahara filtering, bilateral filtering, isotropic TV model filtering, isotropic PM model filtering, anisotropy eickert model filtering, wavelet analysis filter), analyzes the characteristics of each filter algorithm, and the 10 kinds of filter algorithm were MATLAB. Will have MATLAB filter to ultrasound mammary gland tumor image filtering, from noise reduction, smooth, details keep tumor defined area, analyzed according to the characteristics and the experimental data algorithm, based on the analysis of partial differential equation model and the spread of the ultrasonic model bilateral filtering tumor image spots (speckle) noise has better inhibition action, and enhance the edge character. Finally through the test every a filter algorithm process after the image, the application of FCM algorithm based on image segmentation, according to tumor area contour segmentation accuracy of the different results, prehensive filter results, based on partial differential equation that the diffusion model and bilateral filtering model is more advantageous to the FCM image segmentation technology segment the precise tumor area. Keyword: Ultrasound image preprocessing tumor。 Based III on the statistical characteristics of the spatial filtering。 Based on the analysis of the resolution spots noise cancellation algorithm IV 目錄 摘 要 ........................................................................................................................................... I Abstract ........................................................................................................................................ II 緒 論 ........................................................................................................................................... 1 研究背景及意義 ............................................................................................................... 1 斑點(diǎn)噪聲介紹 .................................................................................................................. 1 國(guó)內(nèi)外研究現(xiàn)狀 ............................................................................................................... 3 空域?yàn)V波 ............................................................................................................... 3 空與濾波補(bǔ)充說(shuō)明 ................................................................................................. 3 基于偏微分方程濾波 .............................................................................................. 4 多尺度分析能力的小波濾波 ................................................................................... 4 本文的主要工作 ............................................................................................................... 5 第二章 圖像處理中的濾波算法及分析 .......................................................................................... 7 均值 濾波 ....................................................................................................................... 7 簡(jiǎn)述 ....................................................................................................................... 7 算法說(shuō)明： ............................................................................................................. 7 算法分析 ................................................................................................................ 7 中值 濾波 ..........................