基于多种纹理特征的全色图像云雪区特征提取

Feature extraction based on combined textural features from panchromatic cloud and snow region

针对全色图像云检测与雪检测的问题,文中提出了一种基于多种纹理特征的特征提取方法。首先,利用自适应的大津阈值分割算法提取云、雪区域。然后,通过分形维数、灰度共生矩阵、小波变换等方法提取云、雪区域的多种纹理特性。最后,利用径向基核函数的支持向量机(Support Vector Machine,SVM)分类器进行云雪自动检测。典型遥感数据的实验结果验证了本文算法的有效性。

A feature extraction algorithm based on several combined textural features is presented to distinguish cloud and snow from panchromatic images. Firstly, the target region of the image including cloud and snow is picked up by an adaptive threshold segmentation method based on Ostu algorithm. Secondly, different kinds of textural features are extracted from the target region by fractal dimension, Gray Level Co Occurrence Matrix （GLCM） and Discrete Wavelet Transform. At last, SVM classifier with RBF kernel is used to distinguish cloud and snow automatically. Experimental results indicate that the proposed method can obviously improve the accuracy under the typical remote sensing data.