Automatic cross radiation calibration method for wide-field-angle multispectral sensor

Abstract

The invention discloses an automatic cross radiation calibration method for a wide-field-angle multispectral sensor. The method comprises the following steps of respectively simulating the reflectivity of different ground objects on each wave band of a reference sensor and a sensor to be calibrated, and calculating and simulating a reflectivity conversion coefficient by utilizing a multiple linear regression model; according to the reflectivity conversion coefficient and the real surface reflectivity of the reference sensor, obtaining the simulated surface reflectivity of the sensor to be calibrated; and performing BRDF correction on the simulated surface reflectivity of the the sensor to be calibrated to obtain the simulated corrected direction reflectivity of the sensor to be calibrated. The method does not depend on calibration of actual measurement results of a fixed field calibration field so that the method has high-frequency calibration capability. The method does not depend on image pairs with the same observation time and angle so that the calibration work can be automatically carried out aiming at a specific application scene. The method does not depend on the observation of a specific ground target so that samples can be widely selected, and the stability of the radiometric calibration coefficient of the sensor to be calibrated is improved.

Description

technical field [0001] The invention belongs to the field of satellite remote sensing, and in particular relates to an automatic cross-radiation calibration method for a multi-spectral sensor with a wide field of view. Automatic cross-radiometric calibration for medium and high spatial resolution remote sensing satellite data with large field of view. Background technique [0002] The radiometric calibration of remote sensing images refers to converting the digital signal (digital number, DN) obtained by satellite observation into physical radiance (unit: W m -2 ·sr -1 ·μm -1 )the process of. In general, the basic formula for linear scaling is: [0003] L λ = gains·DN+offset (1) [0004] Among them, L refers to the radiance at the entrance pupil of the satellite, λ refers to the wavelength, DN refers to the unprocessed original value of the remote sensing image, gains and offset are the bias and gain, also called the calibration coefficient. The process of linear cali...

AI Technical Summary

Problems solved by technology

However, at present, this method is mainly used for narrow field of view sensors, less for wide field of view sensors

This is because there is an obvious nonlinear distribution of the reflectivity between the wide field of view sensor and the reference sensor. Using a linear relationship model to estimate the reflectivity in a certain direction is prone to large deviations, which will affect the accuracy of the final cross-calibration

In addition, the spatial resolution of the satellite surface BRDF product and the satellite data to be calibrated is quite different, which is easily affected by the problem of "mixed pixels"

Images