SAR image change detection method based on stack semi-supervised adaptive denoising auto-encoder

Abstract

The invention discloses an SAR image change detection method based on a stack semi-supervised adaptive denoising auto-encoder, and aims at solving the problem that in an existing method, the precision of detection on coherent speckle noise points and change areas of many edges is low. The method comprises the steps that a multi-scale difference guide diagram is generated; a first time phase image is taken as the input to train an SDAE; the multi-scale difference guide diagram, the first time phase image and a second time phase image are taken as the input to train the SSADAE, and weights obtained in SDAE training are used in an SSADAE self-adaption error function; the feature vector of the first time phase image and the feature vector of the second time phase image are calculated by the SSADAE; and the feature vector of the first time phase image and the feature vector of the second time phase image are subtracted to obtain a difference vector, an FCM classification is conducted on the difference vector to obtain a change detection result diagram. According to the method, the multi-scale difference guide diagram is proposed firstly and can highlight the change areas in the difference diagram; and the SSADAE proposed later can improve the change detection accuracy by utilizing a small quantity of mark samples in the image.

Description

technical field [0001] The invention belongs to the technical field of image processing and relates to change detection of SAR images, in particular to a SAR image change detection method based on a stack semi-supervised self-adaptive denoising self-encoder. This method can be used in change detection of SAR images. Background technique [0002] Change detection is one of the key technologies in the field of remote sensing. It detects the changes in the gray value or local texture of images in different periods in the same imaging scene, and obtains the shape, position, quantity and other properties of the surface or objects of interest. change information. It has a wide range of applications in the fields of society, environment and military. [0003] In the multi-temporal SAR image change detection method, there are two main routes, one is post-classification comparison (Post Classification Comparison, PCC), and the other is post-comparison classification. The former me...

AI Technical Summary

Problems solved by technology

[0005] In 2012, Maoguo Gong and Yu Cao et al. published the article A Neighborhood-Based Ratio Approach for Change Detection in SAR Images in IEEE Geoscience and Remote Sensing Letters, Volume 9, Issue 2, Page 307-311, which proposed a neighborhood-based ratio operator (NR) , the NR operator adds the heterogeneity / homogeneity operator of the image, but when the noise distribution of the two SAR images before and after the change is inconsistent, the detection effect of this method is not accurate enough

Disadvantages of this method: First, the generation of initial change masks for optical images and SAR images will introduce large errors, and the learning samples with large errors will also have a greater impact on the results; second, SDAE uses an unsupervised method To extract features, so the extracted features have a certain degree of randomness, and the feature change analysis based on the mapping proposed by the author is completely dependent on the features extracted by SDAE, which will further introduce errors

[0008] To sum up, when the image noise distribution before and after the change is inconsistent, the above method is not good enough for edge detail detection, and the overall error rate of change detection is high

Images