Aircraft Target Classification Methods Based On Multifractal And Its Correlation Characteristics In FrFT Domain

Radar target recognition is an important research branch in the field of radar signal processing,which through the detection and processing of radar echoes to achieve the estimation of the types,loads and other information of the radar targets.Low-resolution radars generally do not have high range resolution in the radial and lateral directions due to their inherent system limitations,so it is difficult to reveal the fine structure of the targets.However,it is feasible to use the limited information carried in the radar echoes to roughly classify the aircraft targets.As a tool of signal analysis and data processing,fractional Fourier transform has dual-domain analysis characteristics in time and frequency domains compared with the traditional Fourier transform,which can effectively suppress the clutter and improve the signal-to-noise ratio of the detected signals,so it is widely used in radar signal processing and target detection.Radar background clutter has fractal characteristics,and the existence of targets will change the fractal characteristics of the clutter.Different types of targets have different structures and materials,which will affect the fractal characteristics of background clutter to different degrees,therefore,the fractal characteristics of different types of aircraft echoes are also different.This kind of differences can be extracted as the classification features to realize the classification and recognition of low-resolution radar targets.This paper explores the classification and recognition of low-resolution radar aircraft targets by means of fractional Fourier transform and fractal theory.In the experiments,firstly,fractional Fourier transform is performed on the low-resolution radar target echoes to calculate the optimal fractional Fourier domain.Secondly,we analyze the fractal characteristics of different types of aircraft echoes in the optimal fractional Fourier domain,and further extract their multifractal and multifractal correlation characteristics.Finally,according to the theory of pattern recognition and the extracted features of different types of aircraft targets,we propose the classification and recognition algorithms of low-resolution radar targets based on multifractal and multifractal correlation features.The experimental results show that the target echoes have more obvious multifractal and multifractal correlation characteristics in the optimal fractional Fourier domain than in the time domain.The classification and recognition rates of aircraft targets based on multifractal features in the optimal fractal Fourier domain are higher than that based on multifractal features in the time domain,and the difference between the two algorithms is more than 7%.Similarly,the classification and recognition rates of aircraft targets based on multifractal correlation characteristics in the optimal fractal Fourier domain are higher than that based on multifractal correlation characteristics in the time domain.The classification and recognition rates of aircraft targets based on multifractal correlation features in the optimal fractional Fourier domain are better than that based on the multifractal features in the optimal fractional Fourier domain.