English

应用科学学报

• 通信工程 • 上一篇    下一篇

基于二维混合变换的脂肪肝超声波图像特征提取

肖扬， 张超， 胡绍海   

  1. 北京交通大学 信息科学研究所，北京 100044
  • 收稿日期:2007-11-12 修回日期:2008-04-08 出版日期:2008-07-31 发布日期:2008-07-31

Feature Extraction from Ultrasonic Image of Fatty Liver Using 2D Hybrid Transform


XIAO Yang, ZHANG Chao, HU Shao-hai   

  1. Institute of Information Science, Beijing Jiaotong University, Beijing 100044, China
  • Received:2007-11-12 Revised:2008-04-08 Online:2008-07-31 Published:2008-07-31

摘要：

采用一种二维混合变换: 二维离散傅立叶离散小波混合变换（2-D DFT-DWT），去除与脂肪肝散射粒子混迭的噪声，进而提取脂肪肝散射粒子。该算法从水平方向对超声波图像用DFT进行滤波处理，从垂直水平方向对超声波图像进行DWT尺度纹理信息提取处理。并给出了水平方向的偏DFT的正变换与反变换的定义与算法，及垂直方向的偏DWT的定义与算法。通过实例说明了脂肪肝超声波图像特征提取算法的应用，给出了与二维离散傅立叶变换或二维离散小波变换算法提取图像中的脂肪肝散射粒子纹理的比较结果。

关键词: 二维混合变换, 脂肪肝散射粒子, 超声波图像

Abstract: A hybrid 2D transform, discrete Fourier transform-discrete wavelet transform (2D DFT-DWT), is applied to remove speckle noise mixed in the liver scatters so that liver scatters can be extracted. The proposed feature extraction algorithm uses DFT to filter the ultrasound image along the horizontal direction, and DWT along the vertical direction in order to extract the size feature of fatty liver scatters. The definitions and algorithms of partial DFT and partial DWT are given. Realistic processing of fatty liver ultrasound images are performed to show application of the proposed algorithm, and comparisons with classical 2D DFT and 2D DWT are made.

Key words: 2D hybrid transform, fatty liver scatters, ultrasonic image