Study on geometric correction of airborne multiangular imagery

An automatic image matching algorithm, and its application to geometric correction of airborne multiangular remote sensed imagery, are presented in this paper. The image-matching algorithm is designed to find correct match for images containing localized geometric distortion and spectral variation. Mathematical tools such as wavelet decomposition, B-spline, and multi-variant correlation estimator are integrated in the frame of pyramidal matching. The simulated experiment and our practice in correcting airborne multiangular images show that the matching algorithm is robust to the few random abnormal points and can achieve subpixel match accuracy in most area of the image. After geometric correction and registration, multiangular observations for each ground pixel are extracted and sun/view geometry is also simultaneously derived.