CCD遥感图像几何校正的实时处理算法初探

为向星上处理迈进，实时完成ＣＣＤ遥感图像１Ｂ级几何校正，迫切需要建立一个准确而又能快速进行运算的反映校正前后几何位置对应关系的函数表达式和重采样算法。通过论证把校正前后一维位置对应关系高度精确地表示为一元三次多项式关系，从而改用递推极大地简化计算；在此基础上提出了一维灰度重采样新算法——加权平均法及其快速算法。当整幅图每行均按此处理之后，再根据获得的ＧＰＳ定轨定姿数据加注出图像的地理经纬网，从而得到完整的１Ｂ级几何校正产品。初步微机仿真实验表明，只需适当引入多处理器，本算法达到实时完全可行。

STUDY ON REAL TIME PROCESSING ALGORITHM OF GEOMETRIC CORRECTION FOR CCD REMOTELY SENSED IMAGERY

Since the processing of remote sensing images is increasingly in real time,it is important to propose a new geometric correction algorithm for imagery of linear and area array charge coupled devices(CCD),which can be used to fast and accurately calculate the corrected geometric positions and resample their gray scales.In this paper,the relational expression to shift one dimensional position is proved to be a cubic multinomial,to which fast recurrence can be applied finely,and the one dimensional resample algorithm is also described,both offering great time savings.When every line of an image has been done accordingly ,the lines of longitude and latitude is indicated as well,based on positioning data given by GPS.So far,a rectified image product with complete 1B level is available.Simulation tests with PC have showed that the algorithm is believed to achieve real time capability if multiple processors are introduced appropriately.