利用灰度和纹理特征的SAR图像分类研究

多类别多特征量情况下的合成孔径雷达(SAR)图像的目标分类是一个难以解决的问题.从灰度和纹理模型出发,提出了综合利用灰度和纹理特征的目标分类方法.均值和方差是灰度模型中重要的特征统计量,而能量、熵、对比度、局部相似性和相关性是纹理模型中重要的特征统计量.灰度和纹理特征能确切地描述SAR图像中的目标.通过构造特征向量,定义向量之间的距离,并按照最小距离方法进行目标分类.以一定大小的窗口读入样本,提高了算法的运行速度和抗噪能力.理论上,窗口越大,特征向量值越接近真实值.窗口越小,边缘的分类精度越高.实验表明该方法较好地处理了多类别多特征量情况下的SAR图像分类问题,分类结果是有效的,这为SAR图像目标分类提供了一条简单可行的途径.

SAR image classification using grey and texture features

Target classification of Synthetic Aperture Radar(SAR) images is difficult under multi-class and multi-feature condition.A novel classification algorithm for SAR images is presented by using grey-scale and texture features based on grey-scale and texture models.In grey-scale model,mean and variance are main feature statistical measures;in texture model,the energy,entropy,contrast,local homogeneity and correlation are main feature statistical measures.The grey-scale and texture features may describe the target in SAR image accurately.In this paper,feature vectors are constructed,distances between the vectors are defined,and targets are classified according to minimumdistance method.To further improve computational speed and reduce speckle,a window approach is used in the classification of SAR images.Theoretically,the bigger the window,the more similar the feature vector to the real vector;the smaller the window,the higher precision of edges.Experiments showed that this approach is effective for SAR image classification,which provides a new access to solve the problem of target classification in SAR images.