Polarization SAR Image Change Detection Method Based on Deep Belief Network

Abstract

The polarimetric SAR image change detection method based on deep confidence network includes first inputting two polarimetric SAR images of the same area and different phases to be detected; registering the polarimetric SAR image data of the two phases; Speckle reduction in the image; preliminary manual marking is performed; the polarization coherence matrices TA and TB are obtained respectively from the polarization scattering matrices of the two polarization SAR images; the diagonal elements of the matrices are extracted respectively, and cascaded to form pixel-based features Matrix F; After normalization, the feature matrix F1 is obtained; each element in the feature matrix F1 is taken as a block to form a feature matrix F2 based on the image block; based on F2, the feature matrix D1 of the training data set D and the test data set T are obtained. Feature matrix T1; construct a detection model based on the deep belief network; use the constructed data set to train the detection model; use the trained detection model to detect the image to be detected. The invention has high detection accuracy.

Description

technical field [0001] The invention belongs to the field of image processing, and relates to a polarization SAR image change detection method based on a deep belief network. Background technique [0002] Polarization SAR is a high-resolution active microwave remote sensing imaging radar, which has the advantages of all-weather, all-time, high resolution, side-view imaging, etc., and can obtain richer information on targets. The method of polarimetric SAR image change detection is a method of comparing and analyzing polarimetric SAR imaging of the same place in different periods, and obtaining the change of ground object information in the same geographical location in different periods according to the difference between the information. Polarization SAR change detection has a wide range of applications in military and civilian fields. [0003] Compared with SAR images, polarimetric SAR images contain richer information, which can reveal the scattering mechanism of the tar...

AI Technical Summary

Problems solved by technology

[0005] Since these polarization SAR change detection methods do not take into account the deep feature representation of polarization SAR images, it is difficult to obtain high classification accuracy for polarization SAR images with complex backgrounds.

Images