A coarse-to-fine correction method for seriously oblique remote sensing image

Conventional approaches are unsuitable to effectively correct the image ac quired in the seriously oblique condition which is susceptible to resolution disparity. Considering that the variability of geometric distortion and the control points' distribution play important roles in correction accuracy, this paper introduces a coarse-to-fine procedure for correcting large-angle images. Firstly, the coarse procedure adopts a control points-constrained piecewise polynomial algorithm to geometric correction with control points clustering, where the whole image is partitioned into contiguous subparts which are respectively corrected by different polynomial correction formulae. And in the fine procedure, for compensating the unreliability of the rigid transformation, a nonrigid thin-plate splines model is used for correcting the whole image. The experimental results show that the proposed correction method is significantly outperforming conventional approaches especially in non-flat areas. © 2011 ICIC International.