Radar-optical-topographic transformations for scene content analysis

A general method for transformations from one image to another is described and demonstrated for a typical radar-optical-topographic scene. The technique requires no visual correlation of pixels, although a knowledge of the position and orientation of the radar imaging system is needed. The theoretical concept of optical to radar transformation is explained through a mathematical model of the process, which derives a set of equations from six radar and three optical reference pixels. The technique is implemented by locating the position and orientation information of the chosen radar scene from previously generated data files, and selecting the region of the aerial photography that is assumed to contain the radar image. An error analysis is made, and the calculation error is reduced through software and visual inspection. Due to the ease of use and versatility of the technique, it requires less time than those methods which require a visual correlation of pixels within the two scenes, and is applicable to other types of scene matching.