Zhiyan Liu

Hyperspectral bands reduction based on rough sets and fuzzy C-means clustering

A method of hyperspectral bands reduction based on rough sets and Fuzzy C-Means clustering is proposed, which consists of two steps: first, Fuzzy C-Means (FCM) clustering algorithm is used to classify the original bands into equivalent band groups, which employs the concept of attribute dependency in Rough Sets (RS) to define the distance between a group and the cluster center, viz. the correlatives of adjacent bands; then the data is reduced by selecting only the one with maximum grade of fuzzy membership from each of the groups. So the great number of bands is decreased while preserving most of the wanted information. Simulation results prove the effectiveness of this approach.

Noise reduction for Doppler ultrasound signal based on the adapted local cosine transform and the garrote thresholding method

In this paper, a novel approach, using the adapted local cosine transform combined with the non-negative garrote thresholding, is proposed to remove noise from the Doppler ultrasound signal. In the proposed approach, the local cosine transform is first performed on the signal of interest followed by a search algorithm to select the best basis. Then the coefficients of the obtained best basis are thresholded based on the non-negative garrote thresholding method. By means of the thresholded coefficients of the best basis, the signal is reconstructed. In the simulation study, the estimation precisions of the mean frequency waveform and the spectral width waveform are studied for the signal after denoising. The simulation and clinical results have shown that the proposed approach is superior to ones based on the wavelet transform, especially under low signal-to-noise ratio (SNR) circumstances.

A new adaptive morphological algorithm for contour extraction on ultrasound image

This paper describes a new adaptive algorithm for contour extraction employing fuzzy sets theory together with morphological image processing techniques. Advantage of this algorithm is that its parameter can be adjusted adaptively to extract the best contour. The contour extraction is accomplished in two stages. The first stage employs fuzzy sets theory to enhance image. The second stage extracts the image contour adaptively and effectively employing morphological theory. The validity of the approach is demonstrated using ultrasound image.

An Novel Clutter Rejection Filters Applied to Wideband Blood Flow Velocity Estimation

Color flow imaging plays a major role in the diagnosis of many vascular diseases, especially cardiovascular disease. The imaging quality is highly influenced by the clutter originating from the vascular wall, stationary and slowly moving tissues and relative motion between the probe held by doctor and the blood flow. The clutter rejection filter in front of the mean frequency estimator will restrict the mean frequency estimation range because the power of the signal component induced by slowly moving blood is suppressed as well. Therefore, a novel scheme, using parameter estimation methods based on the two-dimensional correlation function model (2DCM) and the conventional down mixing, is proposed to clutter rejection. With the parameter estimation methods based on the 2DCM, both the center frequency and the mean Doppler frequency are estimated. The two dimensional information in range gate has been also sufficiently used. Simulation results have shown that this adaptive scheme has achieved superior performance in wideband blood flow velocity estimation

Tracking a Maneuvering Target in Clutter by a New Smoothing Particle Filter

Tracking maneuvering target in cluttered environment is a problem of great theoretical and application interest. In this paper a new smoothing particle filter algorithm is proposed which combines the particle filter with a Gibbs sampler to perform the smoothing of the estimation of the target maneuvering and measurement origin. This algorithm is used to estimate states of a maneuvering target from its cluttered measurements, and the simulation results show its powerful ability to solve the problem

Quadrature Doppler Ultrasound Signal Senoising Based on Adaptive Representations

The spectrogram of Doppler ultrasound signal has been widely used in the clinical diagnosis. The additional frequency component arising from noise will produce an adverse effect on the subjective analysis of the spectrogram and its quantitative analysis. Two approaches based on the adaptive representations including the adapted local cosine transform (CPD) and the wavelet packet (WPD), combined with the garrote thresholding, and is proposed to denoising quadrature Doppler ultrasound signal. At first, to avoid the phase distortion induced by denoising the complex Doppler ultrasound signal directly, the directional information is extracted from the quadrature signal. And then the denoising methods based on the adaptive representations are performed on the forward and reverse flow signals, respectively. At last, the estimated signal is reconstructed from the denoised signals using Hilbert transform. The simulation results have shown that these approaches are superior to ones based on the wavelet transform, especially under low SNR circumstances, and moreover, the denoising method using the CPD has achieved the best performance

Data fusion using fuzzy measures and genetic algorithms

This paper proposes an improvement on the fusion method presented previously (1994, 1998). In those methods not only the reliabilities of the sensors are not considered but also the choice of parameter k is relevant to the number of sensors and whether there is opinion close to 0.5. In our method Genetic Algorithms (GA) is used to find the optimal values for the reliabilities of sensors and fuzzy inference rules for determining the parameter k in multi-sensor fusion. Multi-step fusion and one-step fusion methods are formed based on the fusion functions. Simulation results show the effectiveness of the proposed methods.

A Novel Scheme for Suppressing Velocity Ambiguity in Narrowband Blood Flow Velocity Estimation

Color flow imaging plays a major role in the diagnosis of many vascular diseases, especially cardiovascular disease. A novel scheme is proposed to suppress velocity ambiguity induced by the narrowband estimators based on the two-dimensional correlation function model by incorporating with the wideband velocity estimation techniques. The basic idea is that the velocity is first estimated by narrowband estimator based on the two-dimensional correlation function model, and then a group of velocity candidates are formed. From these candidates, the wideband estimators search the true velocity. This scheme is tested by a two-dimensional correlation estimator incorporating with the butterfly search estimation method. Simulation results have shown that this scheme is an effective tool to suppress velocity ambiguity

Noise reduction for quadrature Doppler ultrasound signal based on the adapted local cosine transform

The spectral analysis of the Doppler ultrasound signal based on the spectrogram has been widely used in medicine for the assessment of blood flow in vessels. The additional frequency components arising from noise produces an adverse effect on the subjective study of the spectrogram and its quantitative analysis. A novel approach using the adapted local cosine transform, combined with the non-negative garrote thresholding method, is proposed to remove noise from the quadrature Doppler ultrasound signal. At first, the directional information is extracted from the quadrature signal. Then the denoising method based on the adapted local cosine transform is performed on the forward and reverse flow signals, respectively. At last, the estimated signal is reconstructed from the denoised signals using Hilbert transform. In the simulation study, the estimation precision of the mean frequency waveform and the spectral width waveform are studied for the signal after denoising. The simulation results for the simulated Doppler ultrasound signals have shown that this approach is superior to the one based on the wavelet transform, especially under low SNR conditions.

Noise reduction for Doppler ultrasound signal based on cycle-spinning and thresholding methods

The spectrogram of Doppler ultrasound signal has been widely used in the clinical diagnosis. The additional frequency component arising from noise will produce an adverse effect on the subjective analysis of the spectrogram and its quantitative analysis. The methods combining cycle-spinning with thresholding methods using orthogonal wavelet transform and cosine packet decomposition are proposed to remove noise from Doppler ultrasound signal. At first, a specified transform method is performed on the signal of interest. Then the soft or hard thresholding method is used to shrink the coefficients, from which the denoised signal is reconstructed. In the simulation study, the improvements of the mean frequency waveform and the SNR before and after denoising are studied. Simulation results have shown that the methods incorporating with cycle-spinning are superior to those traditional denoising methods. In view of computational complexity and performance improvements, the orthogonal wavelet transform combined with cycle-spinning is a practical approach for denoising Doppler ultrasound signal.