Estimating Camera Position and Orientation from Geographical Map and Mountain Image

We describe a method for recovering the camera position and orientation parameters from a single mountain image. It is based on the alignment of the mountain skyline with a synthetic skyline generated from a digital elevation map. Image plane alignment of three image skyline feature points with three model feature points is rst hypothesized and the position and orientation parameters for the hypothesis are computed by nonlinear least squares optimization. The hypothesis is then veri ed by checking for skyline alignment. Search space is reduced using a strategy presented in this paper. Simple skyline and feature point extraction procedures are also discussed.