Factor analysis model-based high resolution range profile self-adaptive framing method

Abstract

The invention discloses a factor analysis model-based high resolution range profile self-adaptive framing method. The method includes the following steps that: the normalized spectrum amplitude feature of an actually measured high resolution range profile signal sample is extracted, the spectrum amplitude feature is evenly divided into small sub-frames; corresponding factor analysis models are established separately, the parameters of the models are calculated through using an EM iterative algorithm, so that the probability distribution of the sub-frames can be obtained; similarity measurement is performed on the sub-frame models through using the JS (Jensen-Shannon divergence) in the statistics, and adjacent sub-frames with high similarities are merged; and the operation of the second step and the third step is performed circularly until the Jensen-Shannon divergences between all the sub-frames are larger than a set threshold value, a frame merging process is terminated, and a framing result is outputted. According to the method of the invention, on the basis of the factor analysis models, and the Jensen-Shannon divergence is used to complete the merging of the frames; and as indicated by an experiment, the method of the invention enables a better framing effect and has higher recognition precision compared with a traditional uniform framing method and a newer probability principal component analysis model.

Description

technical field [0001] The invention belongs to the technical field of signal recognition, and relates to a method for framing one-dimensional radar signal recognition, in particular to a method for adaptively framing one-dimensional range images based on a factor analysis model, and in particular to a method for framing through a factor analysis model and JS Optimal Framing Method Combining Distance Measures. Background technique [0002] High-resolution one-dimensional range profile (HRRP) is a kind of target reflection signal based on broadband radar detection. It is the projection vector sum of target scattering point echoes in the direction of radar line of sight. It contains important structural information such as the size and shape of the target and Compared with SAR and ISAR images, it is easy to acquire and process, so it is widely used in the field of radar automatic target recognition, and it is the most promising radar target recognition method. [0003] Howeve...

AI Technical Summary

Problems solved by technology

However, this method is unreasonable. In the process of the target attitude changing continuously, the speed of change should directly affect the result of framing, that is to say, the frame size should be adaptively adjusted by the speed of the target attitude change.

For adaptive framing, predecessors also proposed a variety of methods, such as framing based on signal similarity, HRRP framing method based on subspace model, framing method based on probabilistic principal component analysis model and KL distance, etc. These The methods have completed the adaptive adjustment process of frame division with the change of target attitude to a certain extent, and achieved good results in recognition rate, but they all have certain defects, that is, the noise of HRRP is not well considered, and there is no Considering the existence of noise or assuming that the noise variance is fixed, this does not match the reality and affects the accuracy of actual framing

Images