Extraction method and recognition method of discriminative feature fusion of multi-pulse radar signal

Abstract

The present application relates to a fusion extraction method and recognition method of distinguishing features of multi-pulse radar signals. The method includes: performing maximum value pooling and average value pooling on the multi-pulse radar signal feature map in the spatial dimension to obtain the first channel weight coefficient and the second channel weight coefficient; according to the first channel weight coefficient, the second channel weight coefficient The channel weight coefficient, activation function and multi-pulse radar signal feature map are obtained to obtain the channel weight feature map, and the maximum pooling and average pooling are respectively performed on the channel dimension to obtain the feature map with the number of channels being 2; according to the number of channels 2 feature map and channel weight feature map to obtain a high-resolution feature map, add it to the multi-pulse radar signal feature map to obtain a high-resolution feature map with complete information, and use multiple convolution kernels for feature fusion extraction , to obtain high-resolution features. The method can quickly and accurately extract high-resolution features in multi-pulse radar signals.

Description

technical field [0001] The present application relates to the technical field of radar signal recognition, in particular to a fusion extraction method and a recognition method for distinguishing features of multi-pulse radar signals. Background technique [0002] With the rapid development of digital technology, the modulation methods of radar signals are becoming more and more complex, and there are more and more modulation types. The external electromagnetic environment is becoming more and more complex, which brings severe challenges to electronic reconnaissance and electronic countermeasure systems. In the process of electronic countermeasures, the rapid and accurate identification of intercepted signals can give priority to the right to control information, which plays a vital role in the battlefield situation. However, the intercepted enemy signals are not only single-pulse signals, so how to quickly and accurately identify multi-pulse radar signals under low signal-to...

AI Technical Summary

Problems solved by technology

However, since the two-dimensional time-frequency images of different radar signals have large repeated similar areas, while the distinctive feature areas are relatively small, the ability to identify multi-pulse radar signals using convolutional neural networks is poor, and the amount of parameters and calculations are relatively large

Images