In-Sik Choi

Efficient radar target classification using adaptive joint time-frequency processing

This paper presents a new target recognition scheme via adaptive Gaussian representation, which uses adaptive joint time-frequency processing techniques. The feature extraction stage of the proposed scheme utilizes the geometrical moments of the adaptivity spectrogram. For this purpose, we have derived exact and closed form expressions of geometrical moments of the adaptive spectrogram in the time, frequency, and joint time-frequency domains. Features obtained by this method can provide substantial savings of computational resources, preserving as much essential information for classifying targets as possible. Next, a principal component analysis is used to further reduce the dimension of feature space, and the resulting feature vectors are passed to the classifier stage based on the multilayer perceptron neural network. To demonstrate the performance of the proposed scheme, various thin-wire targets are identified. The results show that the proposed technique has a significant potential for use in target recognition.

Natural frequency-based neural network approach to radar target recognition

Time domain response-based neural networks and frequency domain response-based neural networks have been proposed for radar target recognition. We propose a natural frequency-based neural network for radar target recognition. Our scheme takes advantage of an aspect angle independence of a natural frequency. It is shown from experimental results that a natural frequency based-neural network using the first natural frequency pair is superior to a time domain response-based neural network in the case of a single aspect angle and that a natural frequency based-neural network using the first natural frequency pair or the first two natural frequency pairs is superior to a time domain response-based neural network in the case of a multiple aspect angle.

Radar Target Recognition Using One-Dimensional Evolutionary Programming–Based Clean

In this work, we present a method for radar target recognition using an one-dimensional (1-D) evolutionary programmingbased CLEAN. This 1-D scattering center extraction method relies upon evolutionary programming and an undamped exponential model. It is accurate, robust and fast. Moreover, it is free from the resolution problems that arise in FFT-based CLEAN. Unlike with model-based techniques, the accuracy of extracted parameters is unaffected by false estimation of the number of scattering centers. Experimental results show that the 1-D evolutionary programming-based CLEAN algorithm can be successfully applied to radar target recognition with correlation-based approaches to reduce data storage, as well as with neural network-based approaches to efficiently extract feature vectors.

Natural frequency extraction using late-time evolutionary programming-based CLEAN

In this paper, we present a novel method for natural frequency extraction. Our algorithm is called late-time evolutionary programming (EP)-based CLEAN, and has many advantages compared to conventional methods. The accuracy of our algorithm is not affected by the false estimation of the number of natural resonance modes. Furthermore, our method is insensitive to random noise. Insensitivity is a very important characteristic in the resonance extraction algorithm since the late-time response usually has small energy. Using synthetic data, we show these characteristics by comparing them to Pronys method and the E-pulse technique. We also applied our method to the numerical data and B-52 measured data which is obtained at Michigan State University (MSU) arch range.

Wide-Angle Radar Target Recognition with Subclass Concept

The range profile is an easily obtainable and promising feature vector for a real-time radar target recognition system. However, the range profile is highly dependent on the aspect angle of a target. This dependency makes the recognition over a wide angular region difficult. In this paper, we propose a classifier with a subclass concept in order to solve this dependency problem. Recognition results with six aircraft models measured at a compact range facility are presented to show the effectiveness of the proposed classifier over a wide-angular region.

Optimized Design of Radar Absorbing Materials for Complex Targets

In this paper, we present a hybrid technique for designing RAM optimally to reduce the RCS of complex targets in a wide-band frequency range. The technique combines a high-frequency method and a genetic algorithm (GA) to obtain an optimal RAM in complex targets. By the virtue of the high-frequency method, such as the physical optics (PO) method and the method of equivalent currents (MEC), the proposed technique can be applied to complex targets with relative ease. However, the high-frequency method needs a classification of shadow regions as pre-processing. A Z-buffer algorithm is employed in this process. The procedure results in designing the optimal RAM which significantly reduces the RCS of complex targets.

Performance enhancement of target recognition using feature vector fusion of monostatic and bistatic radar

This paper proposes a fusion technique of feature vectors that improves the performance of radar target recognition. The proposed method utilizes more information than simple monostatic or bistatic (single receiver) algorithms by combining extracted feature vectors from multiple (two or three) receivers. In order to verify the performance of the proposed method, we use the calculated monostatic and bistatic RCS of three full-scale aircraft and the measured monotatic and bistatic RCS of four scale- model targets. The scattering centers are extracted using one-dimensional FFT-based CLEAN and then used as feature vectors for a neural network classifler. The results show that our method has better performance than algorithms that solely use monostatic or bistatic data.

Determination of optimum bistatic angle for radar target identification

The transmitter and receiver positions of a bistatic radar are highly influential on its performance in radar target identification since the radar cross-section of a target varies with these positions. In this study, radar target identification performance using calculated bistatic scattering data for three full-scale models and measured data for four-scale-model targets is analyzed and compared. FFT-based CLEAN is used for shift-invariant feature extraction from the bistatic scattering data of each target, and a multilayered perceptron neural network is used as a classifier. The optimum receiver position is found by comparing the calculated identification probabilities while changing the position of the bistatic radar receiver. The identification results using calculated data and measured data show that an optimally positioned bistatic radar yields better identification results, demonstrating the importance of the positions of the transmitter and receiver for bistatic radar.

A Study on the Target Recognition Using Bistatic Measured Radar Signals

This paper shows the research about radar target recognition using the measured radar signals from MSU(Michgan State University) bistatic radar system. In this research, we first did the bistatic measurements at , , using F-14, Mig-29, and F-22 scale models. Then, we extract the target feature vectors using time-frequency analysis methods such as STFT(Short Time Fourier Transform) and CWT(Continous Wavelet Transform) and perform the target classification test using MLP(Multi-layerd Perceptron) neural network. The results show that the target classification performance is too much dependent on the bistatic angles and the best performance is obtained at the bistatic angle.

Performance Comparison of PSO-Based CLEAN and EP-Based CLEAN for Scattering Center Extraction

Particle swarm optimization (PSO) is a new high-performance optimizer that can be easily implemented. We investigated the performance of PSO-based CLEAN and EP-based CLEAN for extracting target scattering centers. Simulation results using artificial and measured data show that PSO-based CLEAN is faster than EP-based CLEAN without degradation of accuracy.