Technologies for high-precision automatic georeferencing and high-level processing of spaceborne SAR images

Abstract

Georeferencing quality is one of the main characteristics of the Earth’s surface images generated by spaceborne synthetic aperture radars. The paper considers the problem of automatic high-precision georeferencing of spaceborne synthetic aperture radar images. Three options for solving this problem are considered: by focusing the radio holograms with account of the calibration data which contain corrections to the radar antenna attitude measurements; by autofocusing of the holograms, specifying the Doppler centroid values, and estimating the slant range time scale shift; by correlation-extreme matching of radar and reference images, specifying the radar antenna phase center trajectory. Reducing the georeferencing error to a value comparable to or not exceeding the size of the image pixel projection onto the earth’s surface makes it possible to significantly simplify high-level processing of multi-temporal synthetic aperture radar image sets. The issues of radar images’ matching based on a strict geometric model in their high-level processing and radar and optical images’ correlation-extreme matching in their integration are also considered. The results of experimental research in terms of assessing radar and optical images’ matching accuracy are presented. Visual inspection and the obtained upper limit of the numerical error estimate confirm the pixel accuracy of the matching.

Keywords

• radar space mage
• synthetic aperture radar
• georeferencing
• high-level processing
• matching