Digital geometric picture correction using a piecewise zero-order transformation

This paper describes a procedure for the digital geometric registration of digitized air photographic data. Match points are located visually and used to formulate a global linear conformal transformation for each slave picture. Each transformation then serves to segment the corresponding image into regions such that, for the implementation of the correction, all pixels within a region can be shifted by the same number of rows and columns. In other words, the actual correction is achieved by a series of local translations (zero-order transformations). Remaining problems are then the filling of holes in the output pictures and the finding of a common submatrix. The image registration which results, provides a preprocessing operation needed to combine data from multitemporal and multispectral photography.