| The Directional Polarimetric Camera(DPC)with multi-angle,multi-spectral,and polarization detection capabilities will play an essential role in aerosol inversion and other fields.High-precision geolocation and multi-dimensional data registration are the keys to ensuring the validity of the DPC’s remote sensing data,as crucial as high-precision radiation and polarization detection performance.High-precision geometric calibration is an effective means to ensure the geolocation accuracy and registration performance of multi-dimensional remote sensing data.A reasonable imaging geometric model is the basis for high-precision geometric calibration.In this paper,the imaging geometric model of the DPC is constructed by combining the basic imaging geometric principles of spaceborne optical instruments and the actual instrument characteristics of the DPC.The results of the Zemax analysis show that the radial distortion characteristics of the instrument can be accurately characterized by using the 9th-order distortion coefficient.In addition,this paper systematically discusses the on-orbit geolocation algorithm of the DPC and the registration method of data between multi-angle,multi-spectral,and polarization channels.Geolocation and multi-angle,multi-spectral registration accuracy depend on high-precision imaging geometric models.The registration between polarization channels is realized in hardware based on the optical wedge and is not affected by the imaging geometric model of the instrument.Laboratory geometric calibration is an essential part of the research on the geometric calibration of spaceborne optical instruments.High-precision laboratory geometric calibration can not only systematically and comprehensively evaluate the geometric performance of the instrument,but also verify the accuracy of the imaging geometric model.The high-precision laboratory geometric calibration results will be used for the geolocation and multi-dimensional data registration of the instrument in the early stage of on-orbit operation,and provide a reasonable initial value for the research on on-orbit geometric calibration.The laboratory geometric calibration of the DPC is divided into two parts:the calibration of the internal parameters of the instrument(distortion center coordinates,distortion coefficients)and the calibration of the installation angle.The laboratory uses a geometric calibration method based on a separated two-dimensional turntable and a single collimator to calibrate the internal parameters of the instrument.Two main errors in this method need to be corrected:1)the centroid location error caused by the relative response difference of the entire image plane of the instrument,and 2)the centroid-image point approximation error caused by using the centroid coordinates to approximate the image point coordinates.This paper proposes a centroid location error correction method based on the relative response coefficient and a centroid-image point approximation error model to correct the above errors.The DPC’s laboratory geometric calibration experiment verified the effects of the two correction methods.The correction amount of centroid location error can reach 0.1 pixels,and the correction amount of centroid-image point approximation error can reach 0.7 pixels.Finally,based on the optimized method,the high-precision calibration of the internal parameters of the DPC was achieved in the laboratory,and the model fitting residual error was better than 0.2 pixels.In addition,this paper measured the installation angle of the DPC in the satellite body coordinate system based on the reference prism in the laboratory.Factors such as the violent vibration during the launch of the satellite and the difference between the on-orbit environment and the laboratory environment cause a certain degree of difference between the on-orbit geometric performance of the instrument and the laboratory.To ensure the accuracy of the DPC’s on-orbit geolocation and multi-dimensional data registration,it is necessary to study a high-precision on-orbit geometric calibration method.First,through simulation analysis,this study determined that the parameters that need to be calibrated on orbit in the imaging geometric model are the distortion center coordinates,distortion coefficient,and installation angle.The large width of the DPC makes the traditional on-orbit geometric calibration method based on high-precision ground reference data unable to be effectively applied to the on-orbit geometric calibration of the DPC.This paper proposes an on-orbit autonomous geometric calibration method that does not require ground control point information and is only based on the consistency constraints of the geolocation of homologous points between multi-angle images of the instrument.And the matching error of the homologous point matching method based on SIFT(Scale Invariant Feature Transform)algorithm in the on-orbit autonomous geometric calibration method is systematically analyzed.The on-orbit geometric calibration experiments show that the average residuals of the geographic coordinates of the homologous points in the 865 nm band are reduced from-1.570 km,1.417 km,and 0.701 km to 0.005 km,-0.016 km,and 0.003 km,respectively.This proves the effectiveness of the on-orbit autonomous geometric calibration method proposed in this paper.Finally,this paper systematically evaluates the DPC’s on-orbit geolocation accuracy and multi-angle,multi-spectral,polarization data registration performance.After the on-orbit geometric calibration,the significant improvement in geolocation accuracy,multi-angle data registration performance,and multi-spectral data registration performance further verify the effectiveness of the on-orbit autonomous geometric calibration method in this paper.Taking the 910 nm band as an example,the average geolocation error is reduced from 3.222 km before on-orbit geometric calibration to 1.504 km after on-orbit geometric calibration,and the consistency of geolocation accuracy in different regions on the DPC image plane and in different regions of the world is significantly improved.After on-orbit geometric calibration,the average registration accuracy of the DPC’s multi-angle data is better than 1.530 km,and the average registration accuracy of multi-spectral data is better than 0.650 km.The image registration error between polarized channels increases with the increase of the field of view,the maximum average registration error is about 0.7 pixels,and the average registration error in the 30° field of view is better than 0.1 pixels.In addition,this paper also evaluates the temporal evolution characteristics of geolocation accuracy,multi-angle data registration performance,and multi-spectral data registration performance,which do not show obvious trends over time within the evaluation time range.In this paper,through the in-depth study of the laboratory geometric calibration method and the on-orbit geometric calibration method of the DPC,the geolocation accuracy and multi-dimensional data registration performance of the instrument are improved.This research is of great significance for improving the quality of the DPC remote sensing data and the performance of inversion products and also provides a reference for the geometric calibration research of the spaceborne large field of view instruments. |
PDF Full Text Request