Design Of Software For Vegetation Analysis Based On Hyperspectral Remote Sensing Images

Vegetation as an important part of nature, has a great affect on energy balance , climate change and biochemic circle. Using hyperspectral remote sensing images to analyze vegetation always play an important role in application of hyperspectral remote sensing images. It can overcome the shortcoming of traditional technique ,which use human to analyze vegetation on the spot. Hyperspectral remote sensing images also make it possible to get vegetation information of great area in a wink. So it is necessary and signeficative to analyze vegetation using hyperspectral remote sensing images.The main content of this paper contains design of a software for vegetation analysis .The main contribution are as follow:1. In this paper we find the main reasons to why remote sensing image can't display quickly, and give out our solution to this problem. Based on these reasons ,we define a image format ,which using raster-division technique. We realize the display algorithm using new defined image format. Based on the characters of remote sensing images ,we provide a function of displaying a color image which combines three bands' image .Using these function the user of software can get more information from image.2. In this paper we find that that when we classify a remote sensing image, the pixels show similar spectral curve, but have different pixel value may not belong to the same class. Try to overcome this problem and catch the main difference of different vegetation lays in the difference shape of spectral curve, we put forward a notion called kindred coefficient of direction. Classify the remote sensing images using kindred coefficient of direction, we prove that this coefficient have great advantage of quantifying the difference shape of spectral curve.3. Different vegetation index can reflect different vegetation characters. In this paper we introduce several vegetation indice, their employ situation and their models. Using these models, we realize their calculation. We also introduce the notion of Dimidiate Pixel Model. Based this model, we can calculate the vegetation coverage of any pixel on the remote sensing image.