SAR vehicle target recognition method based on improved convolutional neural network

Abstract

The invention discloses an SAR vehicle target recognition method based on an improved convolutional neural network. The method mainly solves the problem that according to the prior art, accuracy ratesof recognition on SAR vehicle targets are low, and networks are prone to generate over-fitting. The solution of the method includes: removing background clutter of each image in training samples, andcropping each SAR image; constructing an improved convolutional neural network structure based on a caffe architecture, namely setting a classifier of a target recognition part of the convolutional neural network to mixed maximum boundary softmax; inputting the cropped training samples into the improved convolutional neural network to carry out training to obtain a trained network model; carryingout background clutter removal and cropping operations on test samples; and inputting the processed test samples into the trained improved convolutional neural network model to carry out a test to obtain a recognition rate thereof. According to the method, an accuracy rate of SAR vehicle target recognition is increased, network convergence speed is increased, and generalization performance of thenetwork is improved.

Description

technical field [0001] The invention belongs to the field of radar technology, in particular to a method for identifying radar targets, which is used for identifying ground SAR vehicle targets. Background technique [0002] Synthetic aperture radar (SAR) has the characteristics of all-day, all-weather, long-range, high-resolution, etc., and has played an important role in the fields of reconnaissance, detection and guidance. At present, the automatic target recognition technology ATR based on SAR images, especially the ground target recognition has received extensive attention in various fields. [0003] Convolutional neural network (CNN) is a kind of artificial neural network, which has become a research hotspot in the fields of image recognition and segmentation, speech recognition, and human behavior recognition. Compared with the traditional synthetic aperture radar image automatic target recognition SAR ATR method, CNN can automatically extract image features through t...

AI Technical Summary

Problems solved by technology

The advantage of this method is that by improving the traditional CNN structure, the category discrimination ability of CNN is improved, but this method still does not solve the shortcomings of CNN network that is difficult to converge and is prone to overfitting.

Images