Accurate Positioning of SPOT Image Using Aerial Sequence Imagery

Propose a multi-source image matching method for SPOT 5 and aerial imagery after aero-triangulation to achieve accurate positioning of new SPOT images. Firstly, GCPs are extracted as matching points from GCP database constructed by aero-triangulation results. Secondly, the initial position of matching points on SPOT image is predicted based on original RPC parameters, and preliminary registration of aerial GCPs and SPOT image are achieved by block-by-block rectification. Thirdly cross-correlation matching strategy is carried out to get corresponding points. At last matching errors are eliminated using RANSAC algorithm, and accurate RPC parameters are obtained to achieve accurate positioning of SPOT. The experiments verified the method can automatically achieve precise positioning of SPOT-5 image with existing aerial triangulation results, and no outside work for GCPs is needed.