Abstract
Several studies on geometric correction of LANDSAT MSS imagery have been developed. Those methodologies are classified into three: 1) system correction with satellite data only, 2) black box type correction with ground control points only and 3) combined method with system correction and black box type correction. The third method has been pointed out as the best method in accuracy and computing time. However, this method shall be improved into the one with simultaneous solution since the current method is composed of two steps solution, that is, system correction in the first step and black box type correction in the second step.
The objective of the study is to develop a simultaneous solution by treating the position and attitude of satellite as the unknown variables, which will be determined by ground control points and approximate observed satellite data.
In order to determine those unknown variables, so called colinearity equation which is utilized as the basic principles in photogrammetry, is defined as a function of time. Nineteen unknown variables in total have been determined in this study.
The final goal of geometric correction is to rearrange and resample the image data into the array of map coordinate system. From the results of analysis of influence upon the image distortion resulted from variations of position and attitude of satellite, and scan mirror velocity, the geometric distortion within a subdivided area with 350 pixels and 78 lines can be considered to change linearly with not more than 0.2 pixel's distortion.
In this study the geometric correction including resampling for a scene of LANDSAT MSS imagery was accomplished in about seven minutes by using FACOM M-180 AD (12 MB memory) . This result shows to be better in accuracy and computing time as compared with the results obtained from the existing methods.
