Image sequence analysis for updating a 4-dimensional GIS of coastal dunes

The application problem is to update the digital elevation model of a dune area with remote sensing data taken, with a high degree of overlap, as a video sequence from a light aeroplane. The problem in image analysis is how to estimate geometric parameters of plants and sand volumes from a sequence of images in which the same object is imaged some 128 times from different look-angles. As the classical methods for correspondence analysis of stereo images are not suitable for this purpose, the authors develop a new method based on prediction and parameter estimation. After estimating camera and platform parameters, for each image frame, they estimate the parameters of a piecewise linear object element for each intersect of object with a scanline plane. The method is least squares optimal in the parameter domain over a large set of observations. The method solves most occlusion problems