A multi-source multi-temporal remote sensing image target change monitoring method

Abstract

The invention discloses an object varying monitoring method of a multi-source and multi-temporal remote sensing image. The method includes the following steps: pre-treating multi-temporal and multi-spectral images and full-color images in training images and testing images to obtain average wave band images; extracting features of the average wave band images corresponding with the multi-temporal and multi-spectral images in the training images, conducting study to obtain a varying semantic model; extracting features of the pixels, which have been labeled with object varying types, of the multi-temporal and full-color images in the training images, studying to obtain an object varying type semantic model; extracting features of the average wave band images corresponding with the multi-temporal and multi-spectral images in the testing images, using the varying semantic model to extract an object varying region; extracting features of the object varying area of the multi-temporal and full-color images in the testing images, using the object varying type semantic model to identify object varying types. According to the invention, the method increases detection precision of the object varying region and identification precision of the object varying type, can be widely applied to the fields of urban planning, object varying monitoring, etc..

Description

technical field [0001] The invention relates to the technical fields of remote sensing image processing, feature matching, target change detection, change type identification, etc., in particular to a multi-source multi-temporal remote sensing image target change monitoring method. Background technique [0002] Multi-source multi-temporal remote sensing image target change monitoring is a technology that uses a computer to detect target changes from input images of the same area captured at different times and judge the type of change. Multi-source multi-temporal remote sensing image target change monitoring has a wide range of applications in disaster monitoring, urban planning, target monitoring and so on. However, due to the complexity of multi-temporal remote sensing images and the semantic uncertainty of changes, there are still many challenges in using computers to automatically detect and identify target changes and their change types from multi-temporal remote sensin...

AI Technical Summary

Problems solved by technology

However, these features and classifiers are not strong enough to represent targets in high-resolution remote sensing images, ignoring the semantic differences between changing classes and non-changing classes, and between different target change types, which separates feature extraction and feature classification. The internal connection of the system greatly affects the performance of change detection and change type identification.

Images