On-orbit sar image change detection method based on dap and arelm

Abstract

The invention proposes an on-orbit SAR image change detection method based on DAP and ARELM. The limitation of training data on the detection capability of the method is eliminated. The method comprises the steps that for two rectified SAR image pairs, the logarithmic difference DI is calculated; SLIC ultra-pixel segmentation is carried out on the DI, and the mean and centroid of all super-pixels are acquired; a DAP algorithm is used for ultra-pixel clustering to acquire several classes of super pixel sets; a K-means algorithm is used to classify regions into three classes, wherein the highest set is a strict change region, the lowest set is strict unchanged region, and the remaining is an unknown region; a regional training sample is selected, and the neighborhood features are sent into ARELM for training to acquire the ARELM of trained network parameters; and all pixel neighborhood features to be detected are sent into the trained ARELM for classification, so as to automatically acquire a change detection result graph. According to the invention, the limitation of the training data on the detection capability of the method is eliminated, and the method has the advantages of high detection accuracy, automatic decision and stable and reliable detection capability, and is used for SAR image change detection.

Description

technical field [0001] The invention belongs to the technical field of image detection, and mainly relates to change detection of SAR images, in particular to an on-orbit SAR image change detection method based on DAP clustering and graph regularization-based extreme speed learning machine (ARELM), which is used for SAR image change detection. Change detection. Background technique [0002] Change detection is a technology that analyzes and detects changes in ground objects in an area based on multi-temporal satellite images obtained at different times in the same area. With the continuous advancement of science and technology, change detection technology has achieved remarkable results both in theory and practice, detection methods are becoming more and more mature, and some excellent algorithms have also been applied to change detection technology. Nowadays, change detection technology has been widely used in medical diagnosis, land use, vegetation cover detection and so ...

AI Technical Summary

Problems solved by technology

However, due to the lack of effective information of the samples, the detection accuracy and method stability of the unsupervised SAR image on-orbit change detection method need to be further improved.

[0006] Supervised on-orbit change detection of SAR images generally requires sufficient artificial data samples to train supervised on-orbit change detection methods, which have higher detection accuracy than unsupervised detection methods, but due to these Manually annotated sample information is scarce and expensive, and the parameters of the trained method are fixed, so that when this type of method is faced with data of target areas with different widths, resolutions, and different imaging methods (single-view or four-view), its detection The quality is subject to large fluctuations, which restricts the practical promotion and use of supervised detection methods

[0007] In summary, the detection accuracy of the unsupervised SAR image on-orbit change detection method is insufficient

The supervised SAR image on-orbit change detection method is limited by the demand for a large amount of manually labeled data, and cannot be flexibly applied to SAR data with different imaging methods and resolutions.

In the field of change detection of SAR images, with the continuous improvement of the amount of information contained in the obtained data set, the previous pixel-level change detection methods cannot meet the detection requirements.

Images