Study Of Hyperspectral Feature About Remote Sensing Image

Imaging spectral technique is a new technology developed in 1980s. This technology is able to imagine the same object in the continuous spectral band by integrating the imaging technology and spectral technology, as well as optics, optoelectronics, electronics, information-processing and computer science technology. With wide application of the Imaging Spectrometer, spatial, spectral and time-resolution of the images become higher, which improves the spectrum data enormous. So that, how to use the spectral information of hyperspectral remote sensing image fully is becoming very important.This study focus on the condition that a great deal of spectral information has been not used and it has been lost because few bands are selected. Based on polygon, spectrum of the objects in the image is reconstructed, and the feature of spectral curves have been analyzed and studied. The sort of object is confirmed by kindred of spectral curves. The focus of the study is different from the usual ways of classification. In the usual ways, image is classified firstly, then polygons are drawn out by using various methods. Of course, in order to satisfy the users and accord with the requests of mapping, the polygons must be filtrated about the sawteeth. It is impossible that the spectrum of a pixel is used instead of the spectrum of the whole object, because an object is formed of a large of pixels and the spectrum of a pixel has difference from the other ones. This study has drawn the manual polygons of dissimilar objects firstly. The influence of noisy is weeded by computing the mean of the spectrum about the polygon, and the spectral curve of the polygon is replaced by the spectral corve of mean about the polygon. The every spectral corve of mean about the corresponding polygon is computed, and the comparability of the curves is compared. Accordingly, the sort of object is confirmed. Therefore, operation workload is reduced largely than the classified methods based on pixel. To compare With the traditional classified methods of compress about hyperspectral image data and decrease of dimensions about hyperspectral image data, the information of the spectrum of object has lesser losing. The information of the hyperspectral image data has been used well.