基于梯度方向直方图特征和支持向量机的医学影像分类方法

Classifi cation Method for Medical Images Based on HOG Features and SVM

本文阐述了基于内容的图像检索技术的特点,采用梯度方向直方图(HOG)特征来描述图片的边缘特征,利用支持向量机(SVM)对多类别的图片进行分类,并比较了几种核函数对分类效果的影响,指出基于HOG特征和支持向量机的分类方法可以辅助医护人员进行医学影像检索。

This paper introduces the characteristics of content-based image retrieval technology, and describes the edge features of medical images with histogram of oriented gradient features. Then these images are classified with support vector machine. The effect of several kernel functions on classification results is also discussed in this paper. This paper points out that the classification method based on HOG and SVM can help medical staffretrieve medical imags.