Research On Geometric Calibration Of Linear Array Whisking Broom Thermal Infrared Camera

High resolution thermal infrared remote sensing images can measure the surface temperature at small spatial scales,compared with the traditional low resolution thermal infrared image,high resolution image information has features: significant increase in the amount of information,the target temperature and geometric structure is more obvious.In order to collect large and wide high-resolution thermal infrared images,thermal infrared camera using linear array CCD way to ground imaging.Geometric calibration is the key technology to realize high-precision positioning and high geometric quality image of thermal infrared image,which can guarantee the application of high-resolution thermal infrared image.This paper analyzes the errors introduced by whiskbroom,and points out that the imaging start time,the line integration time,the imaging start angle and the line whiskbroom angular velocity are the key to the geometric calibration of whiskbroom imaging.According to whiskbroom imaging characteristics is equal time interval,imaging start time and line integration time are calibrated,which eliminates the instability of model caused by temporal system inconsistently;The whiskbroom angel segmentation method is used to calibrate imaging start angle and line whiskbroom angular velocity,which overcome the low precision of imaging model caused by whiskbroom angle unevenly.Radiometric calibration uses detector histogram matching method to improve the quality of original thermal infrared images;Thermal infrared images and control images registration uses the least square cross-correlation matching to access control points and calculate the calibration parameters.The first satellite equipped with high spatial resolution thermal infrared camera in China–YG-14 to geometric calibration,by use Anyang area digital orthophoto map and digital elevation model,and verified by Taiyuan area data.The results show that the without GCP plane positioning accuracy is better than 30 pixels,and with GCPs plane positioning accuracy is better than 1 pixels.