Remote sensing aircraft target classification method and device based on double-triple pseudo-twin architecture

Abstract

The invention relates to a remote sensing aircraft target classification method and device based on a double-triple pseudo-twin architecture. The method comprises the steps that a sample set of remote sensing aircraft targets is acquired, the sample set comprises anchor samples, positive samples and negative samples, a remote sensing aircraft target classification network based on the double-triple pseudo-twin architecture is constructed, and the network comprises three classification subnets based on the pseudo-twin architecture; the classification subnet comprises two classification branches formed by a feature extraction network, a full connection layer and a classification layer; a loss function is constructed; and according to the anchor sample, the positive sample, the negative sample and the loss function, the network is trained to obtain a remote sensing aircraft target classification model, and a to-be-detected sample is input into the model to obtain a remote sensing aircraft target classification result. According to the method, the capability of distinguishing non-homologous characteristics is improved by adopting a contrast loss improvement model; the triad loss is adopted to shorten the distance of the same kind of targets in the feature space and increase the distance of the non-same kind of targets in the feature space, so that the classification precision is improved.

Description

technical field [0001] The present application relates to the technical field of image processing, and in particular to a remote sensing aircraft target classification method and device based on a double triplet pseudo-twin architecture. Background technique [0002] With the rapid development of remote sensing technology, the resolution of remote sensing images has gradually increased, and the spatial and texture information of images has become more and more abundant. Fine recognition of remote sensing aircraft targets is a hot research topic in the field of remote sensing in recent years. Its main process is to classify according to the target features extracted by the network. How to effectively extract target feature information and be able to distinguish different types of targets with small differences is the key to remote sensing aircraft target recognition. Due to the similar size and shape of different types of aircraft targets, the inter-category feature differe...

AI Technical Summary

Problems solved by technology

[0004] The existing classification model based on the twin architecture has achieved relatively advanced performance, but the recognition accuracy for fine-grained is not high

Images