Joint feature extraction and classification method based on adaptive chirp wavelet filtering

Abstract

The invention provides a joint feature extraction and classification method based on adaptive chirp wavelet filtering, and the method comprises the following steps: S1: carrying out off-line trainingan adaptive chirp wavelet filter; and S2: performing feature extraction and classification on a target signal by using the adaptive chirp wavelet filter obtained in S1, and outputting an identification result. The method integrates a neural network structure with an adaptive filter, takes chirp wavelet atoms as a feature extraction basis function of an input layer of the filter to realize featureextraction of signals, and an output layer of the filter realizes target classification through an adaptive filtering algorithm. The invention improves the accuracy and the anti-noise performance of target identification; the filter parameters can be dynamically adjusted, and the filter effect on unknown signals is more efficient; the joint feature extraction and classification method based on adaptive chirp wavelet filtering has better adaptability to the random identification process; and the parameter adjustment can be carried out in real time according to different target characteristics and identification environments, and the identification performance of an identification system is improved.

Description

technical field [0001] This creation involves the fields of signal processing and pattern recognition, especially a joint feature extraction and classification method based on adaptive chirp wavelet filtering. Background technique [0002] With the rapid development of artificial intelligence technology and signal processing technology, automatic target recognition technology based on radar high-resolution range image has broader application prospects and more urgent application requirements. The radar high-resolution range image is a distribution map of the scattering characteristics of the radar echo signal along the range direction, which provides a rapid and effective target description method. Because the radar high-resolution range image has the advantages of easy acquisition, simple calculation, and ability to reflect the geometric size and structural characteristics of the target, many scientific research institutions have devoted themselves to the research of automa...

AI Technical Summary

Problems solved by technology

[0005] First of all, due to the complex scattering characteristics of the high-resolution range image, the stable scattering model of the moving target cannot be accurately obtained, and there are scattering point shadowing phenomena, which make it more difficult to extract effective features of the target;

[0006] Secondly, because the target recognition in the real environment is particularly complex, the environmental noise often causes great interference to the recognition system, thus affecting the target classification results;

[0007] Third, due to the complexity of the target features and the uncertainty of the recognition process, it is often difficult to obtain the optimal solution for the feature extraction and classification parts of the recognition system at the same time, resulting in poor performance of the recognition system;

[0008] In the prior art, due to the above-mentioned problems, misjudgment and misrecognition may occur, thereby causing heavy property losses, and even causing casualties and accidents, etc.

Images