Tai-Chiu Hsung

The Hong Kong Polytechnic University, Hong Kong, Hong Kong

Are you Tai-Chiu Hsung?

Claim your profile

Publications (49)50.18 Total impact

  • [show abstract] [hide abstract]
    ABSTRACT: A reliable speech presence probability (SPP) estimator is important to many frequency domain speech enhancement algorithms. It is known that a good estimate of SPP can be obtained by having a smooth a-posteriori signal to noise ratio (SNR) function, which can be achieved by reducing the noise variance when estimating the speech power spectrum. Recently, the wavelet denoising with multitaper spectrum (MTS) estimation technique was suggested for such purpose. However, traditional approaches directly make use of the wavelet shrinkage denoiser which has not been fully optimized for denoising the MTS of noisy speech signals. In this paper, we firstly propose a two-stage wavelet denoising algorithm for estimating the speech power spectrum. First, we apply the wavelet transform to the periodogram of a noisy speech signal. Using the resulting wavelet coefficients, an oracle is developed to indicate the approximate locations of the noise floor in the periodogram. Second, we make use of the oracle developed in stage 1 to selectively remove the wavelet coefficients of the noise floor in the log MTS of the noisy speech. The wavelet coefficients that remained are then used to reconstruct a denoised MTS and in turn generate a smooth a-posteriori SNR function. To adapt to the enhanced a-posteriori SNR function, we further propose a new method to estimate the generalized likelihood ratio (GLR), which is an essential parameter for SPP estimation. Simulation results show that the new SPP estimator outperforms the traditional approaches and enables an improvement in both the quality and intelligibility of the enhanced speeches.
    Digital Signal Processing 12/2012; 22(6):1161–1173. · 1.92 Impact Factor
  • [show abstract] [hide abstract]
    ABSTRACT: A reliable estimator for speech presence probability (SPP) can significantly improve the performance of many speech enhancement algorithms. Previous work showed that a good SPP estimator can be obtained by using a smooth a-posteriori signal to noise ratio (SNR) function, which can be achieved by reducing the noise variance when estimating the speech power spectrum. In this paper, a wavelet based denoising algorithm is proposed for such purpose. We first apply the wavelet transform to the periodogram of a noisy speech signal to generate an oracle for indicating the locations of the noise floor in the periodogram. We then make use of that oracle to selectively remove the wavelet coefficients of the noise floor in the log multitaper spectrum (MTS) of the noisy speech. The remaining wavelet coefficients are then used to reconstruct a denoised MTS and in turn generate a smooth a-posteriori SNR function. Simulation results show that the new SPP estimator outperforms the traditional approaches and enables a significantly improvement in the quality and intelligibility of the enhanced speeches.
    Circuits and Systems (ISCAS), 2012 IEEE International Symposium on; 01/2012
  • [show abstract] [hide abstract]
    ABSTRACT: In optical phase shift profilometry (PSP), parallel fringe patterns are projected onto an object and the deformed fringes are captured using a digital camera. It is of particular interest in real time three-dimensional (3D) modeling applications because it enables 3D reconstruction using just a few image captures. When using this approach in a real life environment, however, the noise in the captured images can greatly affect the quality of the reconstructed 3D model. In this paper, a new image enhancement algorithm based on the oriented two-dimenional dual-tree complex wavelet transform (DT-CWT) is proposed for denoising the captured fringe images. The proposed algorithm makes use of the special analytic property of DT-CWT to obtain a sparse representation of the fringe image. Based on the sparse representation, a new iterative regularization procedure is applied for enhancing the noisy fringe image. The new approach introduces an additional preprocessing step to improve the initial guess of the iterative algorithm. Compared with the traditional image enhancement techniques, the proposed algorithm achieves a further improvement of 7.2 dB on average in the signal-to-noise ratio (SNR). When applying the proposed algorithm to optical PSP, the new approach enables the reconstruction of 3D models with improved accuracy from 6 to 20 dB in the SNR over the traditional approaches if the fringe images are noisy.
    Applied Optics 07/2011; 50(21):3973-86. · 1.69 Impact Factor
  • [show abstract] [hide abstract]
    ABSTRACT: Traditional Fourier transform profilometry (FTP) removes the zero spectrum of the acquired fringe image using the linear filtering approach. Such approach assumes there is no aliasing between the zero spectrum and the higher harmonics of the image, which however is not true in general. Thus it cannot adapt to sharp illuminant changes in the fringe image. One practical solution is to exploit one more fringe pattern with π phase shift to eliminate the zero spectrum. This however complicates the system in that the projection of the fringe patterns should be synchronized with the capturing device. Motion or imperfect synchronization makes the subsequent D reconstruction erroneous. In this paper, we introduce a 2D joint bilateral filter to help in removing the zero spectrum. Since the bilateral filter is sensitive to abrupt changes in the fringe image, we can accurately extract the zero spectrum even if there are sharp spatial variations in the pixel magnitude. Experimental results show that the proposed algorithm is superior to the traditional methods and facilitate accurate reconstruction of objects' 3D model.
    01/2011;
  • Source
    Tai-Chiu Hsung, D.P.K. Lun, Wan-Chi Siu
    [show abstract] [hide abstract]
    ABSTRACT: This paper presents the reply for the comment made on "The Discrete Periodic Radon Transform" by A. M. Grigoryan. This comment presents a series of paper about tensor and paired transform as applied to the fast realization of two-dimensional discrete Fourier transform. The reply indicates that the claim made by A.M Grigoryan was incorrect. The contributions of T. Hsung and D. P. K. Lun are far more than that described by A.M Grigoryan.The DPRT is not only a forward transform for computing 2-D DFT. It is a complete discrete transformation method that includes efficient inverse transform.
    IEEE Transactions on Signal Processing 12/2010; · 2.81 Impact Factor
  • Source
    Tai-Chiu Hsung, D.P.-K. Lun
    [show abstract] [hide abstract]
    ABSTRACT: In optical phase shift profilometry, parallel fringe patterns are projected onto an object and the deformed fringes are captured using a digital camera. It is of particular interest because it enables reconstruction of the 3D shape of the object using just a few image captures, which facilitates real time applications. However, when using the approach in real life environment, it is noticed that the noise in the captured images can greatly affect the reconstruction quality. In this paper, we firstly analyze why the noisy fringe images can best be analyzed using the oriented 2D dual tree complex wavelet transform. We then suggest an effective yet simple method for enhancing the noisy fringe images. Both the simulation and experiment results show that the new approach can give good performance in reconstruction with fringe images even at high noise level.
    Image Processing (ICIP), 2010 17th IEEE International Conference on; 10/2010
  • Daniel Pak-Kong Lun, Tai-Chiu Hsung
    [show abstract] [hide abstract]
    ABSTRACT: To obtain a reliable estimate of the a-priori signal to noise (SNR) ratio is crucial to most frequency domain speech enhancement algorithms. Recently, the low variance multitaper spectrum (MTS) estimator with wavelet denoising was suggested for the estimation of the a-priori SNR However, traditional approach directly plugs in the wavelet shrinkage denoiser and adopts the universal threshold which is not fully optimized to the characteristic of the MTS of noisy signals. In this paper, a two-stage estimation algorithm is proposed. First, the log MTS components that are dominated by noise are detected and removed in the wavelet domain. Second, a modified SUREshrink scheme is applied to further remove the noise remained in the speech spectral peaks. The new estimator is applied to the traditional Wiener filter and log MMSE speech enhancement algorithms and leads to significantly better performance.
    International Symposium on Circuits and Systems (ISCAS 2010), May 30 - June 2, 2010, Paris, France; 01/2010
  • [show abstract] [hide abstract]
    ABSTRACT: This paper presents a speech enhancement method based on an improved wavelet based multitaper spectrum (MTS) a-priori signal to noise (SNR) ratio estimator and an harmonics regeneration noise reduction technique. The major advantage of this method is that we can effectively suppress noise in a noisy speech even in very low SNR environments without sacrificing speech intelligibility. In this paper, we devise a speech enhancement system using the improved wavelet based algorithm for the estimation of the a-priori SNR of noisy speeches, and the harmonics regeneration technique is applied to further restore the harmonics and remove the noise remained in the speed spectral. Objective evaluation results show that the proposed method demonstrates better noise suppression ability over conventional approaches without evident degradation of speech quality.
    01/2010;
  • [show abstract] [hide abstract]
    ABSTRACT: Advanced motion tracking systems have been generally applied to movie production and digital entertainment. Due to the successful of Nintendo's Wii game machines, low cost motion tracking solutions are of particularly great demand in the computer gaming industry. In this paper, a low cost optical based motion capture system is proposed. The proposed system makes use of 2 high frame rate DLP projectors which project specially designed IR light patterns to the target object with IR sensing device attached. The pattern of illumination as recorded by the IR sensing device is converted to digital data words and sent via a wireless channel back to the central computing unit for estimating the 3D coordinates of the target object. The system has the advantage of low cost with competitive performance in the update rate of the system.
    01/2010;
  • Tai-Chiu Hsung, D.P.-K. Lun
    [show abstract] [hide abstract]
    ABSTRACT: Classical speech enhancement algorithms often require a good estimation of the short-time power spectrum using, for instance, the periodogram methods. However, it is well known that traditional periodogram methods are prone to induce large variance, hence produces the "musical noise" after enhancement. To alleviate this problem, multitaper spectrum (MTS) estimators with wavelet denoising were proposed. In this paper, we investigate the properties of the MTS of noisy speech signals. We find that, in the log MTS domain, the variance of noise varies according to the magnitude of the underlying speech spectrum. It implies that when applying wavelet denoising to the log MTS, the constant threshold used in the traditional methods is not appropriate. Based on this observation, we further develop a wavelet denoising method with adaptive threshold for estimating power spectrum using multitaper. Simulation results show that the spectrum estimated using the proposed method is consistently more accurate than the traditional uniform thresholding methods. Hence, it further improves the current speech enhancement algorithms using the MTS approaches.
    Circuits and Systems, 2008. ISCAS 2008. IEEE International Symposium on; 06/2008
  • Tai-Chiu Hsung, D.P.-K. Lun, Yu-Hing Shum, K.C. Ho
    [show abstract] [hide abstract]
    ABSTRACT: Prefilters are generally applied to the discrete multiwavelet transform (DMWT) for processing scalar signals. To fully utilize the benefit offered by DMWT, it is important to have the prefilter designed appropriately so as to preserve the important properties of multiwavelets. To this end, we have recently shown that it is possible to have the prefilter designed to be maximally decimated, yet preserve the linear phase and orthogonal properties as well as the approximation power of multiwavelets. However, such design requires the point of symmetry of each channel of the prefilter to match with the scaling functions of the target multiwavelet system. It can be very difficult to find a compatible filter bank structure; and in some cases, such filter structure simply does not exist, e.g., for multiwavelets of multiplicity 2. In this paper, we suggest a new DMWT structure in which the prefilter is combined with the first stage of DMWT. The advantage of the new structure is twofold. First, since the prefiltering stage is embedded into DMWT, the computational complexity can be greatly reduced. Experimental results show that an over 20% saving in arithmetic operations can be achieved comparing with the traditional DMWT realizations. Second, the new structure provides additional design freedom that allows the resulting prefilters to be maximally decimated, orthogonal and symmetric even for multiwavelets of low multiplicity. We evaluated the new DMWT structure in terms of computational complexity, energy compaction ratio as well as the compression performance when applying to a VQ based image coding system. Satisfactory results are obtained in all cases comparing with the traditional approaches.
    IEEE Transactions on Signal Processing 01/2008; · 2.81 Impact Factor
  • Source
    Int. J. Image Graphics. 01/2008; 8:109-134.
  • C.Y. Ma, Tai-Chiu Hsung, D.P.-K. Lun, K.C. Ho, H.K. Kwan
    [show abstract] [hide abstract]
    ABSTRACT: The generalized sidelobe canceller (GSC) efficiently realizes the optimal linearly constrained minimum variance (LCMV) beamformer and delivers excellent beamforming results by reducing directional interference and noise. However, when the input signals are contaminated by other type of noises, such as background and diffused noises, the overall performance of GSC can be even worse than the traditional delay-and-sum beamformers. In this paper, we investigate the application of the spatially adaptive multiwavelet (MWT) denoising technique to the GSC in an environment with severe diffused noise. Comparing with the traditional scalar wavelets, the multiwavelets can better characterize the noise information in the signal such that better denoising performance can be achieved. Different approaches for integrating the GSC and the multiwavelet denoiser were studied. It is found that by adding two denoisers, one to the fixed constrained part and another to the final GSC output, an improvement of 2dB in SNR can be achieved as compared with the traditional GSC method
    Circuits and Systems, 2007. ISCAS 2007. IEEE International Symposium on; 06/2007
  • Chim-Yal Ma, D. P.-K. Lun, Tai-Chiu Hsung
    [show abstract] [hide abstract]
    ABSTRACT: One of the major problems of the discrete multiwavelet transform is its high computational complexity. It stems from the use of the matrix-based multifilters and the extra pre and postfiltering stages, which are popularly adopted for initializing the input data. Recently, we proposed a generalized structure for the discrete multiwavelet transform (GDMWT). The new structure allows the prefilters to embed into the transform such mat the related complexity of implementing the prefilters can be reduced. In this paper, we further consider the efficient realization method of GDMWT. We show that the matrix-based multifilters can be simplified by applying different Givens rotations to the factorized matrices. A new lattice structure is then proposed for their efficient realizations. As compared with the scalar wavelet transform for example using biorthogonal 9/7 wavelet, the proposed approach has a similar complexity while enjoying a better energy compaction performance.
    01/2007;
  • Source
    Tai-Chiu Hsung, D.P.-K. Lun
    [show abstract] [hide abstract]
    ABSTRACT: Prefilters are generally applied to the discrete multiwavelet transform (DMWT) for processing scalar signals. To fully utilize the benefit given by the multiwavelets, we have recently shown a maximally decimated orthogonal prefilter which preserves the linear phase property and the approximation power of the multiwavelets. However, such design requires the point of symmetry of each channel of the prefilter to match with the scaling functions of the target multiwavelet system. A compatible filter bank structure can be very difficult to find or simply does not exist, e.g. for multiplicity 2 multiwavelets. In this paper, we suggest a new DMWT structure in which the prefilter is combined with the first stage of DMWT. The advantage of the new structure is twofold: First, the computational complexity can be greatly reduced. Second, additional design freedom allows maximally decimated, orthogonal and symmetric prefilters even for low multiplicity. We evaluated the computational complexity and energy compaction capability of the new DMWT structure. Satisfactory results are obtained in comparing with the traditional approaches
    Image Processing, 2006 IEEE International Conference on; 11/2006
  • Source
    Tai-Chiu Hsung, D.P.-K. Lun, K.C. Ho
    [show abstract] [hide abstract]
    ABSTRACT: Traditional design of critically sampled prefilters for discrete multiwavelet transform ignores the preservation of the linear phase property, which is important for many applications, such as image coding and digital communications. Balanced multiwavelets solve this problem but make the filters longer. By using linear phase filter banks, we propose a simple algorithm for the design of orthogonal symmetric prefilter banks that can be used with the discrete multiwavelet transform. The prefilter bank resulted is orthogonal and critically sampled and can preserve the approximation power of the multiwavelet as well as the linear phase property. Experimental results show that the systems using the proposed symmetric prefilter banks give better performance as compared with using nonlinear phase prefilters.
    IEEE Signal Processing Letters 04/2006; · 1.67 Impact Factor
  • Tai-Chiu Hsung, Yu-Hing Shum, D.P.-K. Lun
    [show abstract] [hide abstract]
    ABSTRACT: Traditional design of critically sampled prefilters for discrete multiwavelet transform in scalar signal processing ignores the preservation of the linear phase property, which is very desirable for many applications such as image coding and digital communications. Balanced multiwavelets solve this problem by making the filters longer. By using linear phase filter banks, we propose a simple design algorithm of orthogonal symmetric prefilter banks for discrete multiwavelet that are critically sampled and can preserve approximation power. This keeps the overall system as compact as possible. The resulted systems give better performance as compared with using non-linear phase prefilters.
    Intelligent Signal Processing and Communication Systems, 2005. ISPACS 2005. Proceedings of 2005 International Symposium on; 01/2006
  • [show abstract] [hide abstract]
    ABSTRACT: Prefiltering is required to initialize the scalar signal for the discrete multiwavelet transform (DMWT). Traditional prefllters are not linear phase, often nonorthogonal or nonmaximally decimated. Recently, we suggested an orthogonal symmetric prefilter bank (OSPFB) to tackle this problem but it was still difficult to find a compatible filter bank structure and sometimes there was none, especially for low multiplicity multiwavelets. A new generalized DMWT structure was then proposed by combining the prefllters with the first stage of DMWT. It has the advantages of lower computational complexity and more freedom on designing the prefllters. In this paper, the performance of the new DMWT structure is evaluated by applying to a VQ-based image coding system whose codebook is initialized with a modified approach called cumulative absolute difference (CAD). Satisfactory results are obtained in all cases comparing with the traditional approaches
    01/2006;
  • [show abstract] [hide abstract]
    ABSTRACT: Resource availability-based admission control schemes usually offer a binary accept or reject decision on an individual video stream to a video streaming proxy server. An acceptance is granted whenever the server can guarantee the availability of resource required to support the stream. Otherwise, it will reject the stream. This is the greedy approach. However, our study has discovered that an acceptance of some streams may jeopardize the capacity of a video streaming proxy server. In other words, a server will have its capacity greatly reduced after accepting the requests of some streams "black sheep". If we are able to screen out such streams in advance, we can maintain server capacity by handling them more carefully. In this paper, we have developed several schemes to facilitate such diversion and have determined their strength and weakness in different scenarios under a multiple server platform. To avoid the laborious screening process, we have employed a heuristic function as an indicator to measure the loading status of the server.
    VLSI Design and Video Technology, 2005. Proceedings of 2005 IEEE International Workshop on; 06/2005
  • Tai-Chiu Hsung, D.P.-K. Lun, K.C. Ho
    [show abstract] [hide abstract]
    ABSTRACT: Denoising methods based on wavelet domain thresholding or shrinkage have been found to be effective. Recent studies reveal that multivariate shrinkage on multiwavelet transform coefficients further improves the traditional wavelet methods. It is because multiwavelet transform, with appropriate initialization, provides better representation of signals so that their difference from noise can be clearly identified. We consider the multiwavelet denoising by using multivariate shrinkage function. We first suggest a simple second-order orthogonal prefilter design method for applying multiwavelet of higher multiplicities. We then study the corresponding thresholds selection using Stein's unbiased risk estimator (SURE) for each resolution level provided that we know the noise structure. Simulation results show that higher multiplicity wavelets usually give better denoising results and the proposed threshold estimator suggests good indication for optimal thresholds.
    IEEE Transactions on Signal Processing 02/2005; · 2.81 Impact Factor

Publication Stats

197 Citations
39 Downloads
50.18 Total Impact Points

Institutions

  • 2002–2012
    • The Hong Kong Polytechnic University
      • Department of Electronic and Information Engineering
      Hong Kong, Hong Kong
  • 1996–2010
    • The University of Hong Kong
      • Department of Electrical and Electronic Engineering
      Hong Kong, Hong Kong