Geometric/radiometric calibration from ordinary images for high resolution satellite systems

This paper presents two techniques respectively devoted to noise and geometric characteristics assessment from standard images instead of dedicated ones. The noise computation technique assumes that high spatial frequencies are sufficiently weakened by MTF so that only noise remains near Nyquist frequency. It uses Fourier Transform or wavelet packet decomposition. The second technique is based upon matching processing between spectral bands assuming the imaging system focal plane has staggered arrays. It yields very accurate information on focal plane layout as well as high frequency attitude disturbances. Results obtained on simulated images as well as Worldview-2 real products are detailed