Automatic land class interpretation method based on low-frequency error elimination

Abstract

The invention provides an automatic land class interpretation method based on low-frequency error elimination, which comprises the following steps of: modeling a low-frequency error of a star sensor into a Fourier series with a known period, and carrying out dimension expansion processing in a UKF (Unscented Kalman Filter) state so as to reduce the influence of the low-frequency error on remote sensing graphic data; obtaining to-be-processed remote sensing graphic data after low-frequency errors are reduced, reading one or more pieces of land parcel information in the remote sensing graphic data, obtaining grid image data corresponding to the land parcel information, extracting image features in the grid image data, and determining the earth surface type corresponding to the land parcel information according to the image features. By eliminating the low-frequency error, the obtained remote sensing graph is more accurate, so that the result obtained by land class interpretation is quicker and more accurate.

Description

technical field [0001] The invention relates to the technical field of remote sensing image processing, in particular to a method for automatic interpretation of terrain types based on low-frequency error elimination. Background technique [0002] In modern earth observation satellites, the accuracy of attitude determination directly affects the quality of remote sensing image processing, and post-event attitude processing can often obtain higher attitude determination accuracy, so it is very important for remote sensing satellites to be able to perform post-event attitude determination with high precision. [0003] The star sensor is the sensor with the highest precision on the star, but due to the influence of precession, nutation, low-frequency error, etc., its accuracy will be greatly affected, and the low-frequency error component can reach tens to tens of arcseconds, and the error It is difficult to eliminate on-orbit, and generally can be eliminated through post-proce...

AI Technical Summary

Problems solved by technology

[0003] The star sensor is the sensor with the highest precision on the star, but due to the influence of precession, nutation, low-frequency error, etc., its accuracy will be greatly affected, and the low-frequency error component can reach tens to tens of arcseconds, and the error Difficult to eliminate on-orbit, generally can be eliminated through post-processing, which greatly increases the burden of uncertainty and post-processing

Images