论文部分内容阅读
多波段光学遥感影像配准是遥感卫星地面预处理中的关键环节,其精度决定了遥感数据产品的质量。一般使用波段间图像灰度相关匹配来实现多波段影像的配准。然而,该类方法的配准精度受地物特征影响较大,海面、港口、沙漠等地物特征不明显存在影像无法配准的现象。本文以北京一号(BJ-1)小卫星多光谱影像的波段配准为研究对象,建立多波段线阵CCD各探元在CCD扫描方向和卫星飞行方向的几何偏差模型,来计算多波段影像间同名像元在图像行和列方向上的位置偏差,进而拟合出一个固定的波段配准模型来实现多波段影像的亚像元配准。该方法已应用于BJ-1和DMC_UK2小卫星多光谱影像的地面预处理波段配准中,配准误差小于0.5像元,具有良好的应用效果。
Multi-band optical remote sensing image registration is the key step in remote sensing satellite ground preprocessing, the precision of which determines the quality of remote sensing data products. Generally use the image gray-scale correlation matching between bands to achieve multi-band image registration. However, the registration accuracy of this kind of method is greatly affected by the features of the features, and the features of the features such as the sea surface, the port and the desert do not obviously have the phenomenon that the images can not be registered. In this paper, the band registration of BJ-1 multi-spectral images of small satellites is studied as the object of study. The geometrical deviation models of each probe of the multi-band linear array CCD in the CCD scanning direction and the satellite flight direction are established to calculate the multi-band image The position deviation of the same name pixel in the image row and column direction, and then fits a fixed band registration model to realize the sub-pixel registration of the multi-band image. The proposed method has been applied to the preprocessing band registration of BJ-1 and DMC_UK2 multi-spectral multi-spectral images. The registration error is less than 0.5 pixel and has good application effect.