基于机器视觉的肝癌超声图像特征信息提取

提出了识别肝癌超声图像的一种特征提取。采用将共生矩阵和多分辨率提取分形特征方法结合，提取空间灰度独立矩阵、灰度差分统计、Laws纹理能舒度量、傅立叶能量谱等特征来实现肝癌超声图像的识别。实验证明，这种有效的特征提取对超声正常肝和肝癌图像具有较高的分类正确率，对缺乏病理专家的医院和远程医疗信息系统中的应用提供了坚实的理论基础。

Extracting Features Information for Ultrasonic Liver Cancer Images Based on Machine Vision

A study about classification of ultrasonic live cancer images is proposed by the texture features, which combines the cooccurrence matrices with Multi-resolution fractal feature. Many features have extracted including the spatial gray level dependence matrices, the gray-level difference statistics, Laws's texture energy measures and the Fourier power spectrum. We have completed to classify normal liver image and ultrasonic cancer live images. The experiment results also show that this study is very effective and supplies a concrete basis to the hospital absence of professional and the telemedicine information system.