A Method To Improve Geometric Correction Accuracy For Multi-Angle Data
Dianzhong Wang,Anmin Fu,Zhiyu Zhang,Yong Pang,Zhifeng Guo
2008-01-01
Abstract:CHRIS (the Compact High Resolution Imaging Spectrometer) on board platform PROBA (Project for On Board Autonomy) is a high spatial resolution, hyperspectral and multi-angle satellite sensor on trial, which can provide a set of images for the same object from five view angles. When ortho-rectifying a image set, this paper first employs a rigorous sensor model- Toutin's model, inputs a set of reported parameters including across track angle, along track angle, IFOV, altitude, period, eccentricity, orbit inclination, started orbit calculation and gives the track information to the raw data. Then the images are ortho-rectified with geocoded images and digital elevation (DEM) data. With a number of sixteen GCPs, it is found the correction precision decreased with view zenith angle, and the RMSE value is improved from one pixel for zenith to about two pixels with 36 degrees off the nadir. As the GCPs are collected carefully, several more GCPs would not improve the situation, which might be due to a variation of initial parameter combinations. Thus we test rational function model (RFM) to improve the precision. When the GCPs are approximately chosen as in Toutin's model, a RFM with three coefficients produces a same precision trend versus view zenith angle except that the RMSE for each angle is well within one pixel. This phenomenon needs further study to be illustrated theoretically.