A 1D Range Profile Recognition Method Based on Adaptive Local Sparse Preserving Projection

Abstract

The invention discloses a one-dimensional range image recognition method based on adaptive local sparse-preserving projection. The method first preprocesses the measured one-dimensional range image signal sample; The local similarity matrix is ​​obtained by the projection method; then the sparse-preserving projection equation, the local-preserving projection equation and the adaptive maximum distance criterion are fused to establish a joint constraint equation group, and the adaptive local sparse-preserving projection matrix is ​​obtained; finally, the training samples and The test sample is projected into a low-dimensional space, and it is trained and classified with a support vector machine. Based on sparse-preserving projection, local-preserving projection and self-adaptive maximum distance criterion, the present invention makes full use of sample sparse reconstruction and identification information contained in neighbor relations combined with self-adaptive maximum distance criterion to extract sample low-dimensional features, and improves the one-dimensional range image signal. High recognition accuracy, reduced feature dimension, and enhanced anti-interference.

Description

technical field [0001] The invention relates to a one-dimensional range image recognition method based on adaptive local sparseness-preserving projection, in particular to a technology for quickly and accurately identifying a radar target one-dimensional range image in an interference environment, and belongs to the technical field of radar one-dimensional signal recognition. Background technique [0002] Radar automatic target recognition is an important research direction in the field of radar signal processing. With the widespread use of radar automatic target recognition technology, people have higher and higher requirements for radar recognition accuracy, real-time performance and anti-interference performance. The amount of data is large in acquisition, storage and application, and the long processing time becomes a major obstacle in its practical process. Radar high-resolution one-dimensional range profile (HRRP), as a one-dimensional signal, is composed of echoes re...

AI Technical Summary

Problems solved by technology

These methods all use the signal itself or its spatial structure characteristics, and integrate it into the extracted low-dimensional features as identification information. Although they can improve the recognition rate and reduce the feature dimension to a certain extent, they have not deeply explored the connection between signals. relationship, the identification information contained in it is also limited, resulting in limited improvement in recognition rate and weak anti-noise ability, it is difficult to achieve satisfactory recognition results in the actual environment

Images