基于改进自适应阈值的肝脏B超图像滤噪方法  

Application of Signal De-noising Method Based on Multi-scale Analysis for B-Scan Ultrasonic Image Recognition of FattyLiver

作　　者：徐立[1] 白金牛[1] 柳原[1] 

机构地区：[1]包头医学院计算机科学与技术系,包头014040

出　　处：《计算机与数字工程》2015年第11期2052-2054,2060,共4页Computer & Digital Engineering

基　　金：内蒙古自治区高等学校科学研究一般项目(编号:NJZY13252)资助

摘　　要：论文通过多尺度分析小波变换,探索一种肝脏B超图像噪声滤除方法,提高脂肪肝B超图像识别的准确率。论文提出一种基于小波变换多尺度分析的改进阈值噪声滤除方法。首先建立散斑噪声模型,然后进行多尺度小波变换,利用改进的自适应阈值滤噪方法去除图像噪声,最后经小波逆变换重构肝脏B超图像。实验结果显示肝脏B超原始图像中的散斑噪声被有效滤除,同时较好地保存了肝脏B超图像的细节信息。该方法可有效地滤除肝脏B超图像中的散斑噪声,较好地保留细节信息,提高了脂肪肝B超图像识别的准确率。A method of B-Scan ultrasonic images denoising based on wavelet transform is explored .It improved accura-cy of B-Scan ultrasonic image recognition of fatty liver .A speckle suppression method for B-Scan ultrasound image based on improved threshold of wavelet transform is presented .Firstly ,a speckle noise model that is carried out to the B-Scan ultra-sound image is built .Secondly ,the image is decomposed into low frequency and high frequency information by wavelet trans-form .Adaptive threshold denoising are carried out to the fatty liver B-Scan ultrasonic images .Finally ,the inverse wavelet transform is processed .The speckle noises in the original image are removed efficiently and the image edge details are re-served .The method is denoising speckle noises and reserving edge details in B-Scan ultrasonic image effciently .It improves accuracy of B-Scan ultrasonic image recognition of fatty liver .

关 键 词：多尺度分析 小波变换 噪声滤除 B超图像 脂肪肝识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]