论文部分内容阅读
卫星载荷研制发射后其光谱和空间观测模式固定,无法根据复杂地表的多样化需求进行实时灵活调整,且目前遥感器波段设置尚不完善还存在优化空间。引进基于蚁群优化算法的波段选择方法(Ant Colony Optimization-based Band Selection,ACOBS),结合北美区域33景AVIRIS航空高光谱图像,开展了不同区域、不同地表覆盖类型的高光谱波段优选研究,发现各地表类型优选波段组合存在一定差异,其中4波段组合中红光、近红外波段为2个共同入选波段,6波段组合中绿光、红光、短波红外波段为3个共有波段,8波段组合中紫光、绿光、红光、红边、近红外1、近红外2、短波红外1、短波红外2为8个共有入选波段,其他入选波段与地表覆盖类型有关。在此基础上,进一步开展了多光谱卫星波段设置评价研究,发现:4波段优化方案中,绿光、红光、近红外波段1(770~895nm)、近红外波段2(900~1 350nm)为最优波段组合;6波段优化方案中,绿、红、红边、近红外1(770~895nm)、近红外2(900~1 350nm)、短波红外1(1 560~1 660nm)为最优波段组合;8波段优化方案中,蓝、绿、红、红边、近红外1(770~895nm)、近红外2(900~1 350nm)、短波红外1(1 560~1 660nm)和短波红外2(2 100~2 300nm)为最优波段组合。研究结果表明Landsat TM/OLI、SPOT等陆地资源遥感器波段设置还存在一定优化调整空间,特别是红边波段在目前传感器波段设置中没有得到足够重视。
After the satellite payload is developed and launched, its spectral and spatial observation modes are fixed, and it is not possible to flexibly adjust in real time according to the diversified needs of complex surface. At present, the installation of the remote sensor band is still imperfect and there is room for optimization. Based on the Ant Colony Optimization-based Band Selection (ACOBS) method and the AVIRIS aerial hyperspectral image of 33 regions in North America, the optimized study of hyperspectral bands in different regions and different types of land cover is introduced. Among the four bands, the middle-red and the near-infrared bands are two common selected bands. In the six-band combination, the green, red and shortwave infrared bands are three common bands, and the combination of the eight bands Medium violet, green, red, red, near infrared 1, near infrared 2, shortwave infrared 1, shortwave infrared 2 for the 8 selected band, the other selected band and the type of surface coverage. Based on this, the multi-spectral satellite band setting evaluation is further studied. It is found that in the 4-band optimization scheme, the green, red and near infrared bands 1 (770-895nm) and the near infrared band 2 (900-135nm) (770 ~ 895nm), near infrared 2 (900 ~ 1 350nm) and shortwave infrared 1 (1 560 ~ 1 660nm) are the most suitable for the 6-band optimization scheme. (770 ~ 895nm), near infrared 2 (900 ~ 1 350nm), shortwave infrared 1 (1 560 ~ 1 660nm) and shortwave Infrared 2 (2 100 ~ 2 300nm) is the optimal band combination. The results show that there is still some room for optimization and adjustment of the remote sensing band of land resources such as Landsat TM / OLI and SPOT. In particular, the red fringe band has not got sufficient attention in the current sensor band setting.