SAR image change detection method based on sparse representation and capsule network

Abstract

The invention provides an SAR image change detection method based on sparse representation and a capsule network, and the method comprises the steps: (1) selecting two multi-temporal SAR images X1 andX2, and obtaining a differential image through a neighborhood logarithm ratio operator; (2) extracting sparse features on the differential image through a sparse representation method, and generatinga feature image; (3) obtaining pseudo tags of initial classification through a fuzzy clustering method FCM, and selecting proper samples from the feature image to make a sample set by adopting a selection principle of high-confidence samples; (4) constructing an improved capsule network, inputting the feature image extracted through sparse representation, and training an optimization network; and(5) testing the network and generating a change detection image. According to the SAR image change detection method, spatial neighborhood information of the SAR images is fully considered, and sparserepresentation is combined with a capsule network, so that the influence of speckle noise is reduced, deep features of the images are extracted, and the precision and speed of SAR image change detection are improved.

Description

technical field [0001] The invention relates to the field of image processing technology, in particular to a SAR image change detection method based on sparse representation and capsule network, which has important research value in the fields of agricultural survey, forest monitoring, natural disaster early warning and the like. Background technique [0002] Synthetic Aperture Radar (SAR) is an active microwave sensor. Its imaging technology uses the principle of synthetic aperture to improve the azimuth resolution, and then captures large-area and high-resolution SAR images. Change detection in remote sensing is to analyze two SAR images acquired at different times in the same geographical area, and identify the changed areas. Since SAR imaging is not affected by external conditions such as light and weather, it can detect ground targets all-weather and in a large area, making the change detection of SAR images have important research significance in agricultural surveys, ...

AI Technical Summary

Problems solved by technology

Although the convolutional neural network has achieved a good detection effect in the SAR image change detection task, the effect of CNN is not very satisfactory for some tilted and rotated objects.

In addition, the convolutional network requires a large amount of data to generalize and use for network learning, which is a major difficulty for the change detection task of SAR images.

Images