Liu Guizhong

Applying support vector machines to voice activity detection

A new voice activity detector (VAD) algorithm using support vector machines (SVM) is proposed in the paper, and the new VAD effectiveness is validated. The sequential minimal optimization (SMO) algorithm for fast training support vector machines is adopted. The proposed VAD algorithm via SVM (SVM-VAD) also uses the characteristic parameters set used by G.729 Annex B (G.729B) VAD. Comparing SVM-VAD with G729B VAD shows that it is effective for applying SVM to VAD. The new proposed VAD algorithm is integrated with G.729B instead of G.729B VAD, informal listening tests show that the integrated speech coding system has a little better efficiency over the G.729B VAD in perceptivity.

Voice activity detection based on short-time energy and noise spectrum adaptation

On the basis of the short-time energy of speech signals and the efficient method of noise statistics adaptation estimation proposed by Sohn et al.(1998), a new highly robust voice activity detection (VAD) rule for any kind of environmental noise is proposed in this paper. The accurate recognition rate of the new method is about five percent higher than that of Sohns method on average, and also has the same merit of tracking the noise spectrum properly as in Sohns method. Simulation experiments show that the new method is an efficient and robust voice activity detector.

Embedded quadtree-based image compression in DCT domain

The success in discrete cosine transform(DCT) image coding is mainly attributed to recognition of the importance of data organization and representation. In this paper, we proposed an embedded image coder based on quadtree set partition in DCT domain (EZDCT) which is suitable for many kinds of DCT coefficients reorganization schemes. The experimental results show that it is among the state-of-the-art DCT-based image coders when compared with the famous DCT-based image coders, such as EZDCT and MRDCT. For example, for the Barbara image, EQDCT outperforms JPEG EZDCT and MRDCT by 3.3,1.71,1.70 dB in peak-signal-to-noise ratio at 0.25 bpp, respectively.The success in discrete cosine transform(DCT) image coding is mainly attributed to recognition of the importance of data organization and representation. In this paper, we proposed an embedded image coder based on quadtree set partition in DCT domain (EZDCT) which is suitable for many kinds of DCT coefficients reorganization schemes. The experimental results show that it is among the state-of-the-art DCT-based image coders when compared with the famous DCT-based image coders, such as EZDCT and MRDCT. For example, for the Barbara image, EQDCT outperforms JPEG EZDCT and MRDCT by 3.3,1.71,1.70 dB in peak-signal-to-noise ratio at 0.25 bpp, respectively.

An efficient error concealment method for JPEG2000 image transmission

This paper presents two error concealment (EC) techniques for image transmission in JPEG2000, one for the lowest frequency coefficients (the low-frequency EC) and the other for high frequency coefficients (the high-frequency EC). The low-frequency EC algorithm uses a data hiding technique and the packet structure of JPEG2000. The low-frequency coefficients, which are taken as the hidden data, are extracted from the compressed bitstream, and embedded back into the same bitstream. The restored hidden data is used to conceal errors. The high-frequency reconstruction is performed on a bitplane basis. The damaged bitplanes are recovered according to the correlation in the wavelet subbands structure, in which the edge information is detected at first. Experiments show the effectiveness of these algorithms.

A new time-frequency analysis based upon AR model

In this paper we analyzed the reasons why the discrete Wigner-Ville-distribution (WVD) of real-valued signal sampled at the Nyquist rate has spectral aliasing, whereas short time Fourier transform (STFT) has not such problems. For the time-frequency resolution of STFT spectrogram is very poor, a novel method of time-frequency analysis based on auto-regressive model (AR) is presented in this paper, which inherits merits of STFT spectrogram and has very good time-frequency resolution. When data for processing are very large, the new method may have excellent performance for promoting velocity of calculating, saving storage and keeping high time-frequency resolution. In addition, the applications of the new method were also illustrated for identifying ripple-fired explosions, the results were compared with that of spectrogram. Experiments showed that the performances of the new algorithm were superior than that of spectrogram.

Construction of a new adaptive wavelet network and its learning algorithm

A new adaptive learning algorithm for constructing and training wavelet networks is proposed based on the time-frequency localization properties of wavelet frames and the adaptive projection algorithm. The exponential convergence of the adaptive projection algorithm in finite-dimensional Hilbert spaces is constructively proved, with exponential decay ratios given with high accuracy. The learning algorithm can sufficiently utilize the time-frequency information contained in the training data, iteratively determines the number of the hidden layer nodes and the weights of wavelet networks, and solves the problem of structure optimization of wavelet networks. The algorithm is simple and efficient, as illustrated by examples of signal representation and denoising.A new adaptive learning algorithm for constructing and training wavelet networks is proposed based on the time-frequency localization properties of wavelet frames and the adaptive projection algorithm. The exponential convergence of the adaptive projection algorithm in finite-dimensional Hilbert spaces is constructively proved, with exponential decay ratios given with high accuracy. The learning algorithm can sufficiently utilize the time-frequency information contained in the training data, iteratively determines the number of the hidden layer nodes and the weights of wavelet networks, and solves the problem of structure optimization of wavelet networks. The algorithm is simple and efficient, as illustrated by examples of signal representation and denoising.

A new iterative algorithm for reconstructing a signal from its dyadic wavelet transform modulus maxima

A new algorithm for reconstructing a signal from its wavelet transform modulus maxima is presented based on an iterative method for solutions to monotone operator equations in Hilbert spaces. The algorithm’s convergence is proved. Numerical simulations for different types of signals are given. The results indicate that compared with Mallat’s alternate projection method, the proposed algorithm is simpler, faster and more effective.

A new improved flexible segmentation algorithm using local cosine transform

To the problem of no overall optimal merger for one-way merger in the segmentation algorithm proposed by Wang et al., (1999), we propose a method of overall optimal search and merger. At the same time, for the problem of merging a segment which has non-value (value-segment) and a segment whose values are zeros entirely (zeros-segment) to a large segment in Wangs method, we also propose a corresponding method to solve the problem. The main techniques use the local cosine transform (LCT) algorithm for a single small segment, rather than folding processing using its original neighboring data, instead of making zero extension, and then fold the each zero-extension segment. A great deal of numerical simulations validate that this new improved technique solves several problems of the binary-based segment algorithm and Wangs segment algorithm; it not only obtains adapted effective segmentation results, but also there are not many redundancy segmentations.

Automatic signal detection based on support vector machine

Algorithm of STA/LTA is frequently used in automatic signal detection, in which the range of detection threshold is (0, ∞), the optimal threshold should be determined by experiment to make a balance between false detection and missing detection. By using the theory of pattern recognition, a new algorithm for automatic signal detection based on support vector machine was proposed and the method of preprocess and pattern feature extraction were discussed as well as the selection of kernel function for support vector machine. The detection performance of the new algorithm was analyzed by means of real seismic data. The experiments showed that the new method could simplify the selection of threshold and detect signal accurately. In addition to the better performance of anti-noise, the ratio of false detection could decrease 85% in comparison with that of STA/LTA.

A new fast friendly window-based congestion control for real-time streaming media transmission

Real-time streaming media over the Internet is an important component of multimedia applications. For the sake of quality of service (QoS), they make rigid demands on bandwidth, delay and packet loss. However, the current Internet does not offer any QoS guarantees to real-time streaming media over it. How to maximize the transmission quality of real-time streaming applications in a best-effort network while friendly sharing bandwidth with non-real time applications like TCP has become an important issue. But now, many real-time streaming applications based on UDP rarely perform congestion control in a TCP-friendly manner, and they do not share the available bandwidth fairly with applications built on TCP. The Internet communication strongly fears that the current evolution could lead to congestion collapse and starvation of TCP traffic. For this reason, TCP-friendly protocols are being developed to behave fairly with respect to coexistent TCP flows. In this paper we present a new window-based congestion control method—fast fair binomial congestion control (FFBCC) for real-time applications. It provides a good performance of bandwidth distribution and TCP-friendliness for real-time streaming transmission while competing bandwidth with TCP flows.