Empirical radiometric correction and atmospheric calibration of multispectral μMCA Snap camera for UAV applications
Abstract
This contribution presents a complex empirical image-based radiometric and atmospheric correction method of multispectral camera Tetracam μMCA Snap with six separated channels. The method is based on investigation of camera radiometric properties in laboratory combined with vicarious atmospheric correction using modified empirical line approach.
A single multispectral image is selected from a field campaign as an example to demonstrate efficiency of proposed method. Sensor linearity was confirmed on each channel with coefficient of regression more than 0.999 regardless of irradiance and exposure time.
Radiometric correction reduced effects of noise, lens vignetting, nonuniform quantum efficiency of Complementary Metal Oxide Semiconductor (CMOS) array and lens distortion. Sensor noise was reduced per-pixel using dark offset images generated in absence of light.
Through repetition, a sensor specific database of average dark offset images was constructed for different exposure times and characteristic of the per-pixel distribution of noise was extracted. Both Vignette effect and nonuniform quantum efficiency effect were corrected per-pixel at the same time from a set of flat field noise corrected images under a uniform illumination using integrating sphere (LabSphere) resulting in an array of correction coefficients.
There was an assumption that the brightest pixel is not affected by any error. Lens distortion was reduced through the implementation of the Brown-Conrady model.
Atmospheric correction was conducted with modified empirical line using calibration target with nine shades of grey to more accurate construction of empirical line.