An Iterative Image/Object Space Matching strategy for determining surface points

In this paper we describe a robust image matching strategy that we have implemented in Intergraph's ImageStation to compute the precise elevation of a point in 3-D space while preserving the XY position. In the preprocessing stage, the correlation surface centered around the point position is examined and a hierarchical vertical line correlation scheme is evoked at the position yielding the maximum correlation coefficient. This result serves as an approximation for the next stage of iterative image/object space matching. To precisely locate the point on the surface, we describe in detail the strategy where the desired elevation is found in an iterative fashion. The sequence begins by computing the matched position in object space followed by projecting the position of the point with the new elevation back to the image for new matching locations. We report results obtained with this strategy and conclude with suggestions on how to use the method in an automatic DEM generation system. -Authors