Weidong Jin

Adaptive Fuzzy Median Filter for Images Corrupted by Impulse Noise

Median filter was once the most popular nonlinear filter for removing impulse noise because of its good denoising power. In this paper, we present a novel adaptive fuzzy median filter to the restoration of salt & pepper impulse noise-corrupted image, which is particularly effective at removing highly impulsive noise. First we estimate noise level based on fussy set theory, then process the corrupted pixel or extend the size of filtering window, at last get the appropriate median value to replace the noisy pixel. The proposed filter has the benefits that it is simple and it assumes no a priori knowledge of specific input image. To demonstrate the capability of our filtering approach, it was tested on several different image enhancement problems. Experimental results have shown that the proposed algorithm works very well not only for images with lower percentage of impulse noises but also with higher percentage of impulse noise.

A novel generalized demodulation approach for multi-component signals

The generalized demodulation method based on maximal overlap discrete wavelet packet transform (MODWPT) is well suitable for separating multi-component signals due to the properties of MODWPT such as decomposing a signal into a sum of components and the same time-resolution in all decomposition layers. However, the related separation-method research meets lots of challenges because of multi-component signals whose components are overlapped in frequency at some instants. In this paper, a novel generalized demodulation approach is proposed to separate each component from the multi-component signals. A generalized short time Fourier transform (GSTFT) is introduced to project the multi-component analytic signal onto the time-frequency plane and a varying window function is used to improve the time-frequency concentration of GSTFT. The time-frequency plane is considered as a 2-D time-frequency image, the singular value decomposition (SVD) and the generalized demodulation technique are applied to obtain the component of interest. Simulation results show that, after using the proposed approach, the time domain waveform and the instantaneous frequency of each component are obtained by analyzing multi-component signals whose components may be well separated in frequency at one instant and overlapped at a later instant. A method to separate multi-component signals.A filter to extract time-frequency track.A GSTFT for time-frequency distribution of signals.

Radar Signal Automatic Classification Based on PCA

This paper introduces an efficient approach to radar signal automatic classification by extracted fusion feature entropy. In this approach, wavelet packet reconstruct coefficient features are extracted from given radar signals in frequency domain based on wavelet packet decomposition. Then, these features are fused with the principal component analysis and a single characteristic feature vector which can effectively represent difference radar signals is obtained. Aiming at the single fusion feature, its energy entropy and symbolization probability entropy are extracted and extracted fusion feature entropy are used to classify emitter radar signal with fuzzy c-mean clustering algorithm. Simulation experiment show that the proposed approach is verified to be highly accurate and robust even in the low SNR, and the classification algorithm needs only very small memory space to store the reference information and can fast implement radar signal classification.

A fast adaptive switching median filter based on measure-integral

Median filter was once the most popular nonlinear filter for removing impulse noise. In this paper, a new fast adaptive switching median filter (FASMF) based on measure-integral is proposed for improving the performance of median-based filters. This efficient filtering technique is implemented as a two pass algorithm. In the first pass, identification of corrupted pixels that are to be filtered is detected into a flag image. In the second pass, a variable window median filter is used to attenuate impulse noise. To improve the speed of median computation, a step function is employed to expand the array for median, then the relationship between median and measure-integral is deduced and an algorithm is gained by it. Experimental results show that our method can not only overcome the constraints in higher level impulse noise filtering but also gain the lower computation cost.

Feature Extraction of Radar Emitter Harmonic Power Constraint Based on Nonlinear Characters of the Amplifier

In radar countermeasures systems, because each emitter has its own electromagnetic properties inside its transmitted signal, the specific emitter can be identified using received radar signals. Traditionally, the specific emitter identification (SEI) depends on analyzing the time-frequency structure within the usage band. In this paper, a new SEI feature extraction approach based on autocorrelation analysis is proposed. To characterize the nonlinearity of the amplifier in the transmitter, the harmonic power constraint characteristic of the output signal of the amplifier is analyzed, and a power series model is applied to describe the output signal. For different amplifiers, harmonic power constraint properties are different. We present the autocorrelation analysis method to estimate each harmonic power spectrum and extract harmonic power ratio signature features from output signal of the amplifier. The validity of the proposed method is confirmed by simulation experiments. Keywords-radar emitter; harmonic power; amplifier nonlinearity; feature extraction

Multi-component LFM radar emitter signal detection based on LWD

Based on local wave decomposition (LWD), a novel approach for detecting the multi-component linear frequency modulated (LFM) radar emitter signal is proposed in this paper. Every complex radar signal is decomposed into its intrinsic mode components, meanwhile the signal local characteristics are dynamically depicted by instantaneous frequencies. The presented approach can successfully abstract the local frequency features and envelop features of every component of the multi-component LFM radar signal, and estimate the numbers and frequency offsets of component signal of the complex radar signal. Theoretical analysis and experimental results indicate that the proposed approach is effective to analyze and detect the multi-component LFM radar emitter signal.

Binary Phase-Coded Sequence Recognition Based on EMD

Identifying the phase-coded sequence of the binary phase-coded radar signal is a difficultly issue in the radar countermeasures. In this paper, a new approach of abstracting waveform features of the radar signal and identifying the phase-coded sequences is proposed, based on experimental model decomposition. The proposed approach can successfully identify complex phase-coded signal. Moreover, the presented approach can be applied as an integral part of radar signal recognition algorithm to prevent the electronic support measure (ESM) system from taking actions against false radar signal recognition and consequently avoid the waste of the available resources. Computer simulation results have shown that the proposed approach can successfully identify phase-coded sequences, and the recognition accuracy is greater than or equal to 89% when the signal-to-noise ratio is as low as 8 dB.