论文部分内容阅读
针对全色CCD光学遥感卫星成像平台,提出了一种基于多属性融合的高分辨率图像云检测方法。首先根据云的物理特性和成像特性展开多属性分析,并提出了基于可分离度的特征选择准则来进行特征参量提取;然后针对两类样本的分布结构差异,结合LDA线性特征压缩算法,削减特征空间的维度;最后利用SVM分类器完成云检测。实验结果表明,该方法适用于高分辨率全色遥感图像的云检测,具有较高的检测概率和较低的虚警概率以及较高的算法执行效率。
Aiming at panchromatic CCD optical remote sensing satellite imaging platform, a high-resolution image cloud detection method based on multi-attribute fusion is proposed. Firstly, multi-attribute analysis is carried out according to the physical and imaging properties of cloud and the feature selection criterion based on the separability is proposed to extract the feature parameters. Then, according to the distribution structure difference of the two types of samples and the LDA linear feature compression algorithm, Space dimension; finally use SVM classifier to complete the cloud detection. Experimental results show that this method is suitable for cloud detection of high resolution panchromatic remote sensing images with high detection probability, low false alarm probability and high algorithm execution efficiency.