Non-Uniformity Correction Method for Splicing Hyperspectral Imager Based on Overlapping Field of View

Hyperspectral remote sensing can provide rich information on the earth's surface, so it has attracted extensive attention from scholars at home and abroad. Affected by the level of technology, a single hyperspectral imager cannot meet the application requirements of a large field of view and high resolution simultaneously. Splicing hyperspectral imager technology combines multiple hyperspectral imagers into one imaging system, effectively expands the field of view of hyperspectral imagers, and is widely used in precision agriculture, earth observation, environmental monitoring, etc. Due to the influence of detector pixel response, optical system, electronic system and other factors, the output of a single pixel of the detector will be inconsistent when the focal plane array of the hyperspectral imager is under the same uniform radiation source, which is the non-uniformity of the hyperspectral imager. The non-uniformity of splicing hyperspectral imager seriously affects the image quality and interpretation.Non-uniformity correction methods are divided into calibration-based and scene-based correction methods. In this paper, non-uniformity models of a single imager and multiple imagers are established based on the analysis of the non-uniformity of the splicing hyperspectral imager developed by Aerospace Information Research Institute. Based on non-uniformity models, a novel non-uniformity correction method based on overlapping fields of view is proposed. This method integrates the laboratory calibration and the real-time flight data. The laboratory radiometric calibration is used to correct non-uniformity of a single image, while the overlapping field of view and wavelet filter isused to correct non-uniformity of multiple imagers. This method only needs to calibrate the radiation of a single imager in the laboratory so that it eliminates the limitation of requiring a large-aperture integrating sphere. Several experiments are carried out to evaluate the quality of images processed by different methods. Images with non-uniformity of two different bands are selected as original images, and original images are processed by the proposed method. In order to quantitatively compare the correction effects of different methods, three evaluation indexes,Improvement Factor (IF), Non-uniformity (NU) and Spectral Angle (SA), are introduced. The results show that the proposed method can correct non-uniformity effectively and preserve the spectrum features simultaneously as much as possible.