The presence of thin cloud in remote sensing images has brought great impact for follow-up image interpretation.An novel algorithm was proposed to remove the thin cloud from remote sensing images based on the theory of High-dimensional space geometrical informatics(HDSGI). One Image is mapped into an original point (vector) of a high-dimensional space by the theory of HDSGI,namely,a remote sensing image including of cloud is an point of high-dimensional space. In this new algorithm,through homomorphic filter,this high-dimensional space point is geometrically decomposed ino two individual sub-spaces of differ ent scales,decomposition components is done "restraining enhancing transformation" separately,the transformed sub-space compo nents had been maken final synthesis be the required remote sensing image without the cloud. Using by actual satellite image and aviation image, experimental results have shown that the algorithm can not only remove the noisy of cloud but also is efficient for enhancing the edge information contrasfing to the homomorphic filter and wavelet transform algorithms.%遥感图像中薄云的存在为遥感图像的判读带来了极大的影响,本文针对遥感图像中的薄云雾提出了一种基于高维空间几何信息学(High-Dimensional Space Geometrical Informatics)(HDSGI)的去云的新算法.基于HDSGI理论,将一副带有薄云的遥感图像作为高维空间中的一点,通过各向同性滤波器将该高维空间点几何分解到不同尺度的两个子空间,对各个子空间的分解分量分别进行抑制、增益变换,将经过变换后的子空间分量合成得到最终所需的去云后遥感图像.采用实际卫星影像和航空影像对该算法进行实验验证,用同态滤波和小波分解算法进行同类算法比较,并采用若干客观图像评价参数评价分析处理后图像.实验结果表明,本文算法不仅能有效降低薄云覆盖及相关的噪音干扰,而且可以增强原始遥感图像边缘信息,达到去除薄云的目的.
展开▼
