A radar high-resolution range profile target recognition method and device

By using a pre-trained HRRP fusion reconstruction model, and utilizing the feature reconstruction and fusion of a variational transformer model and multiple Gaussian distributions, the problem of insufficient expressive power of the Transformer model in complex HRRP sequence recognition is solved, thereby improving the accuracy of radar target recognition.

CN118429683BActive Publication Date: 2026-06-02XIDIAN UNIV

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
XIDIAN UNIV
Filing Date
2024-02-28
Publication Date
2026-06-02

Smart Images

  • Figure CN118429683B_ABST
    Figure CN118429683B_ABST
Patent Text Reader

Abstract

The application discloses a radar high-resolution range profile target recognition method and device, which comprises the following steps: inputting the preprocessed HRRP of a target into a pre-trained HRRP fusion reconstruction model, so as to make the model output HRRP fusion reconstruction features; and identifying the classification of the target according to the HRRP fusion reconstruction features. The feature reconstruction module in the model extracts features from the preprocessed HRRP, and reconstructs the HRRP features according to pre-trained hidden variables subject to multiple Gaussian distributions. The feature fusion module fuses the HRRP reconstruction features according to pre-trained weights, so as to obtain the HRRP fusion reconstruction features. The hidden variables subject to multiple Gaussian distributions can be used to process the preprocessed complex HHRP time sequence, and then the HRRP fusion reconstruction features are obtained by fusing the pre-trained weights, and the classification of the target is identified based on the HRRP fusion reconstruction features, so that the accuracy of the classification and identification of the target is improved.
Need to check novelty before this filing date? Find Prior Art