Dual-temporal remote sensing image change detection method, model construction method and device

Abstract

The invention discloses a dual-time-phase remote sensing image change detection method and a model construction method and device, and belongs to the technical field of remote sensing image processing. A deep residual network model added with an extrusion-excitation module is adopted to construct a dual-time-phase remote sensing image feature extractor; the feature extractor fuses rich semantic information of high-level features and rich detail information of low-level features, an extrusion-excitation module is introduced to weight information of each channel, so that the model pays more attention to important features, the feature extraction effect is improved, and a pyramid attention module is used for detecting different size change targets in the prior art. An image cutting mode is improved, a pyramid attention module cuts a feature image into a plurality of groups of feature sub-images with different sizes and edge pixels overlapped with each other, and the feature sub-images are processed by using a common attention algorithm, so that the detection capability of the model on different size change areas, especially small size change areas, is improved.

Description

technical field [0001] The present invention relates to the technical field of remote sensing image processing, in particular to a dual-temporal remote sensing image change detection method, model building method and device. Background technique [0002] In recent years, remote sensing images have been widely used due to the convenient and accurate acquisition of surface information around the world. As an important branch of remote sensing image research, change detection has been studied for decades. Today, change detection tasks usually include the following aspects: land use change detection, forest and vegetation change detection, urban expansion change detection, disaster monitoring and assessment such as earthquakes and forest fires, etc. Change detection tasks usually involve a vast surface area, and manual execution is time-consuming and laborious. The introduction of change detection algorithms based on deep learning has greatly saved researchers' manpower, materi...

AI Technical Summary

Problems solved by technology

[0008] The present invention provides a dual-temporal remote sensing image change detection method, model building method and device, aiming to solve the problem of the detection effect of the change detection method for dual-temporal remote sensing image in the prior art on targets of different sizes under complex backgrounds. Poor, poor ability to identify changing regions and invariant regions, and the problem of reduced training effect caused by the imbalance of positive and negative samples in the training data set

Images