Synthetic aperture radar anti-deceptive-interference method based on shadow characteristics

Abstract

The invention provides a synthetic aperture radar anti-deceptive-interference method based on shadow characteristics. Firstly, SAR images with and without shadows of several types of targets under different postures are obtained by using a synthetic aperture radar imaging method and an electromagnetic scattering simulation method. The SAR images obtained at different radar incident angles are used as training samples and test samples for a convolutional neural network. In order to overcome the disadvantage of poor shadow characteristic identification effect of the convolutional network, a first-level convolutional neural network is used to classify the targets and the backgrounds to obtain targets and backgrounds of different types, a standard threshold segmentation method and multi-value processing are employed for key target images to obtain multi-value images after target regions are segmented, and real targets and deceptive targets are distinguished by a convolutional neural network classification method. The functions of SAR automatic target identification and interference target identification are realized, and high-performance SAR anti-deceptive-interference in the image domain is achieved.

Description

technical field [0001] The invention belongs to the field of radar technology, in particular to the technical field of synthetic aperture radar (SAR) anti-jamming and synthetic aperture radar (SAR) automatic target recognition (Automatic Target Recognition, ATR) technical field. Background technique [0002] Deceptive jamming achieves the purpose of disrupting the opponent's radar reconnaissance system by simulating the echo signals of false targets or false scenes. As the jammer simulates the SAR echo signal more precisely, the fineness of jamming modulation has been significantly improved, so that the Doppler coherence of the real echo can be simulated more accurately, and the power requirement for the jammer is greatly reduced , and can form more refined deception interference results. International scholars have conducted research on the basic principles of spoofing jamming. Deception jamming has coherence in both the range and azimuth directions. In imaging processing,...

AI Technical Summary

Problems solved by technology

Then, aiming at the shortcomings of convolutional networks that are not effective in recognizing shadow features, the present invention proposes a two-level classification strategy, that is, using the first-level convolutional neural network to classify objects and backgrounds to obtain different types of objects and backgrounds, and then For key targets (such as ground targets, air targets, etc.) images, the standard threshold segmentation method and multi-valued processing are used to obtain the multi-valued image after the target area is segmented. Finally, the convolutional neural network classification method is used for the multi-valued processed samples. , to distinguish the real target from the spoofed target

Images