结合图像特征的支持向量机高分一号云检测

Cloud detection based on support vector machine with image features for GF-1 data

针对高分一号8 m空间分辨率多光谱数据应用中云层影响信息提取精度与影像利用率等问题,提出了一种结合影像光谱特征与纹理特征的支持向量机云检测方法。基于高分一号8 m数据,利用灰度共生矩阵提取其纹理特征,以云与地物的光谱特征和纹理特征构造特征向量,采用支持向量机方法对高分一号数据进行云检测。研究表明,该方法对各类云层检测的查准率与查全率均分别保持在99.2%和93.9%以上,错误率控制在1.1%以下,明显优于利用传统支持向量机与最大似然值法的云检测算法,说明该方法能够很好地检测出影像中的云层,且由于结合了图像纹理及光谱特征,理论上具有一定的普适性。

In the GF-1 image data applications,applying cloud layers influences accuracy of information extraction and image utilization rate,to tackling this problem,this paper proposes a support vector machine cloud detection method combining image spectral features and texture features.For GF-1 data,the method of the gray-level co-occurrence matrix is used to extract those texture features.Spectral characteristics of clouds and ground and texture characteristics serve as feature vector,and support vector machine(SVM)is used to conduct cloud detection to GF-1 data.Studies have shown that the precision and recall of this method for all kinds of cloud detection are above 99.2%and 93.9%,and the error rate is below 1.1%,which is obviously better than the cloud detection algorithm using traditional support vector machine and maximum likelihood value,and it combines the image texture and spectral characteristics,and thus it has certain universality in theory.