Research On Calibration And Hardware Acceleration Technology Of Spaceborne Cloud Detection Camera

With the development of remote sensing technology,the performance of optical remote sensors has been continuously improved,such as spatial resolution,temporal resolution,spectral resolution,etc.And data quality and utilization need to meet higher requirements.However,the mean value of cloud amounts in the world is over 60%.To reduce the impact of clouds on remote sensing data,on-orbit real-time cloud detection technology can be used to detect cloud distribution information and provide pointing information for remote sensors in advance,which can effectively improve the utilization rate of remote sensors.The cloud detection camera is a cloud distribution information acquisition payload designed by the Anhui Institute of Optics and Fine Mechanics,Chinese Academy of Sciences for small satellite platforms.Considering the design requirements and scientific objectives of the instrument,the research work of laboratory calibration and the performance test of the instrument have been carried out.According to the test results,to further improve the performance of the instrument,the hardware acceleration implementation scheme of the instrument is analyzed.Firstly,according to the observation target and index requirements,the system design scheme adopted by the spaceborne cloud detection camera was given.Based on the wide-angle imaging optical system and the area array CCD image sensor,the surface of the pixel of which has the primary color filters arranged by the Bayer rule to achieve color separation.The information transmission model and calibration model of the spaceborne cloud detection camera were established.Secondly,according to the information model and calibration model,the geometric calibration,spectral calibration,and radiometric calibration of the spaceborne cloud detection camera were carried out.In the process of radiometric calibration,it is difficult to perform absolute radiometric calibration for all pixels due to the large field of view of the cloud detection camera and a large number of effective pixels.Therefore,to establish the relationship between all pixels’ response DN values and the radiance,the absolute radiometric calibration was used for the central field of view,and the relative radiometric calibration was combined.The method of sub-field view measurement was designed to conduct the relative radiation response calibration experiment,and a solution was proposed to solve the problem of response crosstalk in the working spectrum of the spaceborne cloud detection camera.Thirdly,the performance test of the spaceborne cloud detection camera was formulated.The hardware performance such as spectral waveband,resolution,width,signal-to-noise ratio and FPGA resource utilization,and the software performance of the on-orbit fast cloud judgment algorithm were tested.And the evaluation results of hardware performance and cloud judgment accuracy were given.Finally,according to the performance test results,the shortcomings of the current cloud detection camera were found that the camera could process fewer pixels in real-time for the observation target in the required time.To improve the speed of cloud judgment,a hardware acceleration solution was proposed.The geometrical relationship between the cloud detection camera and observation targets and the process of on-orbit cloud judgment were optimized locally.Parallel processing was implemented in the hardware circuit by using pipeline,loop unrolling,dataflow optimization and other methods,which greatly improved the calculation speed.