SAR target identification method based on a multi-parameter optimization generative adversarial network

Abstract

The invention discloses a synthetic aperture radar (SAR) target identification method based on a multi-parameter optimization generative adversarial network, and mainly solves the problems that the identification rate is not high during classifier training and the classifier parameters obtained by training cannot be ensured to be an optimal solution in the prior art. According to the implementation scheme, an initial training sample set and a test sample set are generated, and initial training samples are expanded to generate a final training sample set; Setting a structure and a parameter group number of the generative adversarial network; Training the generative adversarial network by adopting a multi-group network parameter cross training method, and training a discriminator in the generative adversarial network by utilizing the training set samples and the pseudo samples generated by the generator at the same time; And identifying the target model by using the trained discriminators in the plurality of groups of generative adversarial networks, adding the results obtained by the plurality of groups of discriminators, and averaging to obtain an identification result of the target model. According to the method, the accuracy of SAR target identification is improved, and the method can be used for identifying the static SAR target.

Description

technical field

[0001] The invention belongs to the technical field of communication, and further relates to a synthetic aperture radar SAR target model identification method, which can be used to identify the model of the stationary target in the synthetic aperture radar SAR target. Background technique

[0002] Synthetic aperture radar (SAR) has the characteristics of all-weather, all-time, high resolution and strong penetrating power. It has become an important means of earth observation and military reconnaissance. Automatic target recognition of synthetic aperture radar (SAR) images has attracted more and more attention. At present, most of the synthetic aperture radar SAR target recognition methods only use the original training data when training the classifier; most of the deep models used in the design of the classifier can obtain local optimal solutions.

[0003] The University of Electronic Science and Technology of China proposed a SAR target recognition method b...

Images