Method for detecting small targets of sea surface by utilizing least squares support vector machine (LS-SVM) on basis of wavelet noise reduction

Abstract

The invention aims to provide a method for detecting small targets of a sea surface by utilizing a least squares support vector machine (LS-SVM) on the basis of wavelet noise reduction. The method comprises the steps that: a section of original sea clutter data is selected, is de-noised and is then subjected to chaos judgment, and embedded dimensions and time delays are extracted; another sectionof original sea clutter data is selected, is subjected to wavelet decomposition and denoising and is normalized; the best regression parameters of the SVM are searched; obtained data are trained by using the LS-SVM as a tool; another group of sea clutter data is selected, is preprocessed and is then predicted by using an obtained sea clutter model, so short-term prediction data corresponding to the data are obtained; the obtained short-term prediction data are subtracted from a data value which is only preprocessed, and absolute values of the difference values are obtained, so as to obtain absolute errors; the absolute errors are classified by using the LS-SVM which is optimized by using particle swarms, two-dimensional information classification is equivalent to threshold limitation, andsmall target detection in a strong sea clutter background is finally realized. By adopting the method, the false-alarm probability is greatly reduced, and thereby the detection performance is improved.

Description

technical field [0001] The invention relates to a target detection method in the technical field of radar signal processing. Background technique [0002] Sea surface target detection (especially small target detection on the sea surface) technology of radar plays a very important role in civilian applications. In various detection methods, it is very important to deal with the noise in the background environment where the target is located, which directly affects the performance of target detection. The difficulty of correctly describing the motion of sea clutter makes small target detection in sea clutter one of the most complex problems in radar signal processing. [0003] Reports relevant to the present invention include: 1. "Research on chaotic time series forecasting based on least squares support vector regression" (the 3rd phase of the 24th volume of "Academy of Naval Aeronautical Engineering" in 2009), which introduces the core theory of support vector machine And...

AI Technical Summary

Problems solved by technology

[0004] Although there have been many studies on the detection of small targets on the sea surface, none of them have been able to provide a complete method for the detection of small targets in the background of sea clutter from the system, which makes it more practical engineering significance

Images