Double-pass SAR image trace detection method based on multivariate statistics and deep learning

Abstract

The invention discloses a double-pass SAR image trace detection method based on multivariate statistics and deep learning. The double-pass SAR image trace detection method comprises the following steps: acquiring a difference image of a double-pass SAR image by utilizing a complex reflection change detection estimator; performing water area and vegetation area identification and false alarm elimination on the difference image by using an unsupervised multivariate statistic method to obtain a false alarm elimination image; training a CUnet network by using inductive transfer learning and images from coarse to fine; and performing trace identification on the to-be-processed image by using the trained CUnet network. According to the method, a difference image of a double-pass SAR image is obtained through a complex reflection change detection estimator, a water area and a vegetation area are obtained through unsupervised multivariate statistics, then a false alarm elimination image is obtained, coarse-to-fine images are constructed by combining an original image, the difference image and the false alarm elimination image, inclusive transfer learning is carried out by using coarse to fine images and CUnet, so that double-pass SAR image trace detection under the condition of small samples is realized, and the detection effect is good.

Description

technical field [0001] The invention belongs to the technical field of target detection, and in particular relates to a method for detecting traces in a double-passage SAR image based on multivariate statistics and deep learning. Background technique [0002] A very important application of the Synthetic Aperture Radar (SAR, Synthetic Aperture Radar) system is to detect footprints, wheel prints and other trace areas, which can be used for surveillance and search purposes. Double flight SAR images are two SAR images obtained by repeatedly flying the same area at different times. Correlation Change Detection (CCD, Coherence Change Detection) has the ability to locate the trace area from a large-scale area, and can be used to realize the trace detection of dual-passage SAR images. The CCD model consists of two modules: a difference generation module, which uses repeated and repeated geometrically registered image pairs to generate a difference image; a difference analysis modu...

AI Technical Summary

Problems solved by technology

Due to changes in the natural environment such as wind blowing and rivers, the difference image generated by the CCD contains a large number of low correlation coefficient areas caused by the natural environment, which brings great challenges to the detection method

Images