The Remote Sensing Research Program University of California at Berkeley has developed a regression-based XY transform methodology and supporting computer software to accurately and cost-effectively transform one- and two-dimensional digital data from one geometrical coordinate system to another. First, control points common to both the original coordinate system and the desired output system are located with respect to each system. Second. the control points and the desired equation form are input to a statistical package program which performs a least-squares regression analysis of the point; to yield transform equation coefficients. Third, the original coordinates of a one-dimensional line segment or a two-dimensional picture element (pixel) are substituted into the transform equations to yield the revised coordinates. This transformation method is computationally inexpensive, can be performed easily on small computer systems, and can be used for many types of remote sensing and ancillary data. Up to four ERTS images have been overlayed with less than one pixel error over a 1024 x 1024 pixel area. ERTS data to Universal Transverse Mercator (UTM) ground coordinate system transforms have been within 200 meter RMS error for the same size area.
Date of this Version