Radar target classification method, device and equipment and readable storage medium

Abstract

The invention discloses a radar target classification method. The method comprises the following steps: acquiring a high-resolution range profile of a radar target; classifying the high-resolution range profile by using a preset convolutional neural network to obtain a classification result of the radar target; and carrying out visual display on the classification result of the radar target. The preset convolutional neural network comprises a convolutional layer, an activation function layer combined with a BN algorithm, a convergence layer and a full connection layer provided with an SVM classifier. The convolutional neural network disclosed by the method can improve the accuracy of the classification result of the radar target and the classification efficiency of the radar target. Correspondingly, a radar target classification device and equipment and a readable storage medium disclosed by the invention also have the above technical effects.

Description

technical field [0001] The present invention relates to the technical field of signal processing, and more specifically, to a radar target classification method, device, equipment and a readable storage medium. Background technique [0002] High Resolution Range Profile (HRRP) is the vector sum of target scattering point echoes projected on the radar ray obtained by wideband radar signal, which provides the distribution of target scattering points along the distance direction, and is an important part of the target. Structure. Therefore, the recognition and classification of radar targets based on high-resolution range images can effectively improve the accuracy of recognition and classification. [0003] In the prior art, when the activation function in the neural network model processing the high-resolution range image of the radar target is processing the activation value, the value will gradually approach the upper and lower endpoints of the value range of the nonlinear...

AI Technical Summary

Problems solved by technology

[0003] In the prior art, when the activation function in the neural network model processing the high-resolution range image of the radar target is processing the activation value, the value will gradually approach the upper and lower endpoints of the value range of the nonlinear function, resulting in a discrepancy between the values. The gradient disappears and it is difficult to converge, which will reduce the efficiency of data processing; and the existing neural network model uses the Softmax function for target classification, and because the Softmax function will consider global samples when performing target classification, resulting in its classification hyperplane is disturbed by each sample, which will reduce the accuracy of the classification result

Images