Clustering Analysis Of Hyperspectral Image And Its Application

Nowadays, Hyperspectral Images have been widely used in variety of fields. Hyperspectral images combine spatial information and spectral information effectively, so they are useful for object detection. With the development of technology, the spatial resolution of hyperspectral images is gradually improving. And hyperspectal images are no longer confined to satellite remote sensing. As the requirement of clustering performance is increasing, the clustering algorithm of hyperspectral images should be improved as well.As we know, cloud plays an important role in atmospheric circulation, and its shape and thickness are the key features in the prediction of weather. So it is very important to identify the cloud in the sky. The technology of hyperspectral image has its own superiority in material identification. So if we can use the clustering of hyperspectral image in cloud identification, the efficiency of the process can be improving.This paper proposes a cluster algorithm based on the density peaks algorithm, which can self-adaptively find the number of clusters and identify the cluster centers. The algorithm is stable, and avoids the process of choosing the initial cluster centers. At the same time, the algorithm proposed by this paper optimizes the calculation of density of data, which simplifies the parameter choosing in the algorithm.This paper uses a LCTF hyperspectral imager to capture a lot of hyperspectral images of cloud. With these images, this paper verifies the function of the new algorithm and compares the result with the traditional cluster algorithm.This paper designs a hyperspectral image clustering system, with which users can analysis the hyperspectral images and clustering with the image data.