TG-2 Multi-Angle Polarization Imager Remote Sensing Image Geolocation And Error Correction

As an important data source of GIS,remote sensing imagery has a clear concept of spatial location and expresses the correspondence between pixels and earth targets.In the data preprocessing stage,it is important as spectral and radiometric calibration,and is closely related to the accuracy of quantitative inversion when generating remote sensing products.The Tiangong-2 Space Laboratory(TG-2),equipped with multiple remote sensing application payloads,was launched in September,2016.The multi-angle polarization imager has 12 observation channels(565nm to 910 nm),which is the first multi-angle polarization instrument used in space exploration in China.In addition to conventional narrow-band radiation measurements,MAI is also capable of measuring the polarized reflectance of three different polarization angles of a channel,and can obtain the reflectance of the target 12 different observation directions in a single pass of the spacecraft.Because the polarization of light has unique sensitivity to the characteristics(shape,size and composition)of atmospheric particles,polarization remote sensing has advantages in target recognition and information extraction that traditional remote sensing does not have.In addition,the addition of multi-angle information data can further improve the accuracy of quantitative inversion.As a payload of the TG-2 Earth observation project,the quality of the data is essential for accurate inversion of atmospheric parameters,however,the multi-angle polarization imager has not been studied in depth since the launch.High-precision image geolocation is the basis for the quantitative application of observation data.In order to generate advanced products,it is necessary to clarify the geographical location of the remote sensing image and convert it into data reflecting multi-angle information.Therefore,it is necessary to fully analyze the accuracy of MAI geolocation,correct the errors occurring during the positioning process,and provide an accurate geometric position of the earth target.Based on the optical geometry design parameters of Multi-angle Polarization imager and the attitude and state information of spacecraft,the relationship model between MAI image observation pixels and ground space position is established by using parameter method,and the geolocation algorithm of the instrument is developed.At the same time,in the process of geolocation error correction,based on the observation mode and error characteristics of the instrument,a method for correcting the error in both along-track and cross-track is proposed.Finally,the image DN value is used as the basic data without radiometric calibration.The coastline crossing method was used to evaluate the geolocation results,and the image registration method was used in combination for cross-validation.The main contents and conclusions of this paper are as follows:(1)Based on the TG-2/MAI optical geometry and instrumental ground observation characteristics,a set of preprocessing algorithm flow from the unpacking of the original measurement data to the determination of the mapping relationship between the image observation unit and the corresponding target position,and then to the multi-angle data reorganization is established,The obtained pixel-by-pixel geometric data with positioning results and additional parameters lay a foundation for subsequent evaluation and correction of MAI remote sensing image geolocation performance and atmospheric parameter inversion.(2)This paper fully analyzes the geometric deformation factors of remote sensing images,such as the Earth's rotation image,terrain fluctuation and atmospheric refraction.Starting from the characteristics of geolocation error,the errors of the three main directions are classified and processed,and the relationship between the rotation deformation of the image around the Z axis and the rotation angle is established by the idea of nonparametric method.According to the error characteristics in both directions of the along and cross,the relationship with the focus of the focal plane is established.and a method for correcting the error of the specific direction is proposed,and the correction effect of the method is analyzed by using existing data evaluation.The results show that the method used in this paper is more simple and convenient,and has the same correction effect.(3)Through the data comparison,the effects of adjusting the mounting matrix and the focus correction method of this paper to correct the errors along the rail and the cross direction are analyzed.It is found that when the mounting matrix is adjusted to correct the errors along-track and cross-track caused by the rotation of the instrument around X and Y axes,the image edges will be stretched and deformed,and the method presented in this paper is effective in correcting the errors in both directions than adjusting the mounting matrix.(4)Based on the error analysis method of the coastline inflection point method,the geolocation error evaluation of the preliminary corrected and final corrected images is performed respectively.The results show that the standard deviations of the geolocation errors after two corrections are larger in the direction along the track than the cross-track,but the magnitude remains basically unchanged.After the final error correction,the average errors in both directions are significantly reduced.Excluding some anomalies,at the position of the nadir,the maximum error along the track and the cross-track direction remains substantially within one pixel.Cross-validation of the final correction results using the image registration method shows that the minimum value of the cost function is obtained when the Suomi NPP shifts 7 and 4 pixels respectively along-track and cross-track direction.