摘要
为了建立适合于脂肪肝B超图象识别的最佳特征矢量,首先提取正常肝和脂肪肝B超图像的近远场灰度比特征以及灰度共生矩阵、灰度游程长度两种纹理模型中的9个特征。再用Kolmogorov-Smirnov假设检验和人工神经网络进行两次特征选择。通过特征选择,最初的10个特征矢量只有近远场灰度比以及灰度共生矩阵的角二阶矩、熵、反差分矩共4个特征被保留下来,用于脂肪肝的识别。这4个特征组成了适合于脂肪肝B超图象分析的最佳特征矢量。研究表明由这4个特征组成的最佳特征矢量对脂肪肝识别有着较好的性能。
During the feature extraction, three models were employed, which include mean intensity ratio of the near field and the far field, the gray level co-occurrence matrices (GLCMs), and the gray level run-length (GLRL). 10 statistics were extracted from the three models for each image. After the feature selection which involves hypothesis tests and artificial neural networks, there are only 4 features left for further researches including the Angular Second Moment (ASM), Entropy (ENT) and Inverse Differential Moment (IDM) from the GLCMs, as well as the Mean Intensity Ratio (MIR). Thus, the best feature vectors which indicate two classes of images are created with the four features. The feature vectors created with ASM, ENT, IDM and MIR have the best performance during the recognizing task.
出处
《四川大学学报（工程科学版）》
EI
CAS
CSCD
北大核心
2005年第1期130-134,共5页
Journal of Sichuan University (Engineering Science Edition)
基金
四川省应用基础研究项目(03JY029 072 2)
关键词
脂肪肝
特征提取
近远场灰度比
灰度共生矩阵
Feature extraction
Image processing
Inverse problems
Neural networks
