高光谱影像预处理技术

 以核工业北京地质研究院遥感重点实验室获取的高光谱数据CASI/SASI的处理为例,总结了高光谱数据处理技术的流程、具体技术细节和技术点,并根据数据处理中存在的问题,例如高光谱数据量大、处理时间长、影像的拼接困难等,阐述了对高光谱影像预处理中所存在问题的相关对策,为高空间分辨率的高光谱数据预处理提供了较好的应用范例和技术支撑.

Preprocessing Techniques for Hyperspectral-images

This paper discusses the processing of the CASI/SASI hyperspectral-image data obtained by Beijing Research Institute of Uranium Geology. The process flow, the technical details and the technical points of the hyperspectral processing are summarized. The problems of the hyperspectral data processing, such as the large hyperspectral data amount, the long processing time and the image stitching difficulties are discussed and the related countermeasures are proposed. A better application model and the technical support are provided for high spatial resolution hyperspectral data. The airborne hyperspectral images would be distorted due to the effect of the terrain, so the high resolution DEM data should be treated by orthorectification. Under different conditions when the ground images are obtained from the equipment on the plane, the same object on two images would have different colors, so the color difference should be eliminated. A great number of tests show that the CROSS model of the ENVI software can eliminate the color difference. As the space and spectral resolutions of the hyperspectral images are improved, the data volume increases. So the problem of the large volume data would add more difficulties and slow the speed of the hyperspectral image disposition. The spare time of the airborne hyperspectral images acquisition procedure should be used to do some work of the hyperspectral image pre-process. The characteristic bands are used to do the re-sampling of the hyperspectral images, which can improve the efficiency and increase the extraction speed of the hyperspectral images.