Remote sensing image change detection method

Abstract

The invention discloses a remote sensing image change detection method using a low-rank prior learning discrimination dictionary, which mainly improves the accuracy of unsupervised remote sensing image change detection and replaces automatic threshold selection with learning of a change discrimination dictionary and a non-change discrimination dictionary. The method comprises the following steps:(1) respectively inputting two multi-temporal remote sensing images to be detected; (2) performing super-pixel segmentation on the difference image to obtain a super-pixel total sample; (3) learning aglobal dictionary and a sparse coefficient for the superpixel total sample; (4) selecting samples according to low-rank representation of the sparse coefficient matrix; (5) respectively learning thedictionaries by using the changed samples and the invariable samples; and (6) calculating a reconstruction error of each pixel point neighborhood block under the discrimination dictionary, and comparing the reconstruction errors to obtain a change region binary image. Change samples and unchanged samples can be selected in an unsupervised manner for change detection, and the method can be appliedto the remote sensing fields of disaster assessment, urban expansion detection, land coverage, utilization detection and the like.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a method for detecting changes in remote sensing images, which can be used in remote sensing fields such as disaster assessment, urban expansion detection, and land cover. Background technique [0002] Change detection is a technique of monitoring the changing area of ​​remote sensing images of the same geographic location obtained from different times. With the improvement of spatial resolution and the expansion of the monitoring range of satellite sensors, how to make full use of the limited spectral information to highlight the change area and improve the accuracy of change detection poses a huge challenge to change detection. [0003] At present, the methods of remote sensing image change detection are mainly divided into three categories, supervised, semi-supervised and unsupervised methods. The main difference between the three categories is whether data labe...

AI Technical Summary

Problems solved by technology

Among them, the method based on deep learning, because it is aimed at the detection of remote sensing images, the data set does not meet the requirements of the training volume, so it cannot get good results in practical applications.

In the supervised change detection method, the accuracy has been greatly improved due to the addition of priority information, but label information is difficult to label; in the semi-supervised change detection method, unlabeled information is introduced to make up for the lack of label information. , but the training samples that are difficult to classify in the training set will weaken the discriminative ability of the learning model

Images