In this paper, a geometry-based image retrieval system is developed for multi-object images. We model both shape and topology of image objects using a structured representation called curvature tree (CT). The hierarchy of the CT reflects the inclusion relationships between the image objects. To facilitate shape-based matching, triangle-area representation (TAR) of each object is stored at the corresponding node in the CT. The similarity between two multi-object images is measured based on the maximum similarity subtree isomorphism (MSSI) between their CTs. For this purpose, we adopt a recursive algorithm to solve the MSSI problem and a very effective dynamic programming algorithm to measure the similarity between the attributed nodes. Our matching scheme agrees with many recent findings in psychology about the human perception of multi-object images. Experiments on a database of 13500 real and synthesized medical images and the MPEG-7 CE-1 database of 1400 shape images have shown the effectiveness of the proposed method.

The essence of fractal image denoising is to predict the fractal code of a noiseless image from its noisy observation. From the predicted fractal code, one can generate an estimate of the original image. We show how well fractal-wavelet denoising predicts parent wavelet subtress of the noiseless image. The performance of various fractal-wavelet denoising schemes (e.g., fixed partitioning, quadtree partitioning) is compared to that of some standard wavelet thresholding methods. We also examine the use of cycle spinning in fractal-based image denoising for the purpose enhancing the denoised estimates. Our experimental results show that these fractal-based image denoising methods are quite competitive with standard wavelet thresholding methods for image denoising. Finally, we compare the performance of the pixel- and wavelet-based fractal denoising schemes.

In this paper, a new 2D shape multiscale triangle-area representation (MTAR) is proposed. This representation utilizes a simple geometric principle, the area of a triangle, in obtaining a robust and efficient shape representation. The use of the wavelet transform for decomposing the boundary of the shapes improves the efficiency and robustness of the representation. The MTAR is more robust to the affine transformation, less affected by noise, and more selective than similar methods, e.g., the curvature scale-space CSS. Two tests, using MPEG-7 CE-shape-1 database, show that MTAR achieves better performance than the CSS under affine transformation and in the general shape retrieval.

A new joint autoregressive (AR) parameter and order estimation approach is presented which works well regardless of noise type. Unlike many other existing AR estimation methods which require prior knowledge of model order and noise statistics, the proposed method provides embedded order estimation in the pole-zero domain without requiring any noise statistics. The joint method virtually achieves the Cramer-Rao bound (CRB) in a general noise environment.

The log energy parameter has long been used as an extension to the basic cepstral feature vector in speech recognition. The use of a normalization technique for the log energy parameter has also been widely accepted. In this paper, a simple energy normalization scheme is introduced that allows direct use of the frame energy parameter in speech recognition and performs well in the presence of noise. Its combination with traditional cepstral mean and variance normalizations has led to error rate improvements of up to 55% on the Aurora 2 task, in comparison to the baseline clean-trained system using feature set including the log energy parameter. This achievement has been obtained with neither complicated programming nor computation expensive routines. The performance of this scheme on an utterance-wide basis has been close to that of the off-line speaker-wide normalization, which makes it a good candidate for practical systems.

Performance of adaptive noise cancellation (ANC) degrades severely when uncorrelated noise components are present at the two inputs. Thus, practical background diffuse noises pose a serious problem for ANC systems. In this letter, we propose a new hybrid system that integrates subband adaptive filters (SAFs) and a Wiener filter. The hybrid system is implemented on an oversampled DFT filterbank that efficiently integrates the SAF and the Wiener filter components in the frequency-domain. Performance evaluation of the hybrid system in presence of diffuse noise interference shows that the proposed system is superior to both the Wiener filter and the SAF subsystems.

The performance of the normalized least mean square (NLMS) algorithm for adaptive filtering is dependent on the spectral flatness of the reference input. Thus, the standard NLMS algorithm does not perform well in over-sampled subband adaptive filters (OS-SAF) because colored subband signals are generated even for white input signals. Thus we propose the use of the affine projection algorithm (APA) to adapt the individual subband filters in OS-SAF systems. The OS-SAF using APA for adaptation is implemented on a fast, low-resource oversampled filterbank. Through both theoretical and experimental analyses, it is demonstrated that a low order APA significantly improves the convergence behavior, offering a low computational complexity compared to the recursive least squares (RLS) method. We employ a recursive method of calculating the correlation matrix to further decrease the computation cost without affecting the performance.

The performance of the normalized least mean square (NLMS) algorithm for adaptive filtering is dependent on the spectral flatness of the reference input. Thus, the standard NLMS algorithm does not perform well in over-sampled subband adaptive filters (OS-SAFs) because colored subband signals are generated even for white input signals. Thus we propose the use of the affine projection algorithm (APA) to adapt the individual subband filters in OS-SAP systems. The OS-SAF using APA for adaptation is implemented on a fast, low-resource over-sampled filterbank. Through both theoretical and experimental analyses, it is demonstrated that a low order APA will significantly improve the convergence behavior, offering a low computational complexity compared to the recursive least squares (RLS) method. We employ a recursive method of calculating the correlation matrix to further decrease the computation cost without affecting the performance.

The convergence rate of the Least Mean Square (LMS) algorithm is dependent on the eigenvalue distribution of the reference input correlation matrix. When adaptive filters are employed in low-delay over-sampled subband structures, colored subband signals considerably decelerate the convergence speed. Here, we propose and implement two promising techniques for improving the convergence rate based on: 1) Spectral emphasis and 2) Decimation of the subband signals. We analyze the effects of the proposed methods based on theoretical relationships between eigenvalue distribution and convergence characteristics. We also propose a combined decimation and spectral emphasis whitening technique that exploits the advantages of both methods to dramatically improve the convergence rate. Moreover, through decimation the combined whitening approach reduces the overall computation cost compared to subband LMS with no preprocessing. Presented theoretical and simulation results confirm the effectiveness of the proposed convergence improvement methods. 1.

Over the past decade, there has been significant interest in fractal coding for the purpose of image compression. However, applications of fractal-based coding to other aspects of image processing have received little attention. We propose a fractal-based method to enhance and restore a noisy image. If the noisy image is simply fractally coded, a significant amount of the noise is suppressed. However, one can go a step further and estimate the fractal code of the original noise-free image from that of the noisy image, based upon a knowledge (or estimate) of the variance of the noise, assumed to be zero-mean, stationary and Gaussian. The resulting fractal code yields a significantly enhanced and restored representation of the original noisy image. The enhancement is consistent with the human visual system where extra smoothing is performed in flat and low activity regions and a lower degree of smoothing is performed near high frequency components, e.g., edges, of the image. We find that, for significant noise variance (sigma > or = 20), the fractal-based scheme yields results that are generally better than those obtained by the Lee filter which uses a localized first order filtering process similar to fractal schemes. We also show that the Lee filter and the fractal method are closely related.

We propose a simple yet effective fractal-wavelet scheme for edge-preserving smoothing of noisy images. Over the past decade, there has been significant interest in fractal coding for the purpose of image compression. Fractal-wavelet transforms were introduced in an effort to reduce the blockiness and computational complexity that are inherent in fractal image compression. Applications of fractal-based coding to other aspects of image processing, however, have received little attention. The authors proposed a simple yet effective fractal-based image denoising scheme that is applied in the spatial domain of the image. We extend the application of this fractal denoising scheme to the wavelet domain of the image. We find that when the wavelet transform of the noisy image is simply fractally coded, a significant amount of the noise is suppressed. However, one can go a step further and estimate the fractal code of the wavelet transform of the original noise-free image from that of the wavelet transform of the noisy image. The use of the quadtree partitioning scheme for the purpose of fractal-wavelet coding results in a significantly enhanced and restored representation of the original noisy image. The enhancement is consistent with the human visual system where extra smoothing is performed in flat and low activity regions and a lower degree of smoothing is performed near high frequency components, e.g. edges, of the image. The main advantage of the wavelet-based fractal denoising scheme over the standard fractal denoising scheme is that it is computationally less expensive.

This paper proposes a scheme of optimal resource management for reverse-link transmissions in multimedia wideband code-division multiple-access (WCDMA) communications. It is to guarantee quality-of-service (QoS) by resource (transmit power and rate) allocation and to achieve high spectral efficiency by base-station assignment. This approach takes the form of a nonlinear-programming large-scale optimization problem: maximizing an abstraction for the profit of a service provider subject to QoS satisfaction. Solutions for both single-cell and multicell systems are investigated. The single-cell solution has the advantage of low complexity and global convergence in comparison with the previous work. Maximum achievable throughput (capacity) of a single cell is mathematically evaluated and used as the benchmark for performance measure of multicell systems. For multicell systems, due to its max-max structure, solving the optimization problem directly entails a high-computational complexity. Instead, the problem is reformulated to a mixed integer nonlinear-programming (MINLP) problem. Then, binary variables indicating base-station assignments are relaxed to their continuous analogs to make a computer solution feasible. Furthermore, approximations can be made to make the resource-management scheme less computationally complex and allow its partial decentralization. The sensitivity of the proposed scheme to path-gain estimation error is studied. Simulation results are presented to demonstrate the performance of the proposed scheme and the throughput improvement achieved by combining resource allocation with base station assignment.

We investigate the practicality of a proposed network-level resource-management scheme for reverse-link transmissions in multimedia wideband code-division multiple-access (WCDMA) communications. The inherently centralized algorithm manages all transmitted powers and rates and base-station assignments to guarantee quality of service while achieving a high spectral efficiency. Approximations make it less computationally complex and allow its partial decentralization. Preliminary simulation experiments and analysis of its sensitivity to path-gain estimation error show reasonable performance. Of interest is its potential combination with closed-loop power control in IMT-2000.

We propose a scheme of optimal resource management for reverse-link transmissions: in multimedia wideband code-division multiple-access (WCDMA) communications. It is to guarantee quality of service (QoS) by resource allocation (transmitted power and rate) and achieve high spectral efficiency by base-station assignment. The approach takes the form of a nonlinear-programming large-scale optimization problem: maximizing an abstraction for the profit of a service provider subject to QoS satisfaction. Due to its max-max structure, solving the optimization problem directly entails a high computational complexity. Instead, the problem is reformulated to: a mixed integer nonlinear-programming (MINLP) problem. Then, binary variables indicating base-station assignments are relaxed to their continuous analogs to make a computer solution feasible. Preliminary simulation results show the effectiveness in a multicell network.

Such codes are described using dual leaf-linked trees: one specifying the parsing of the source symbols into source words, and the other specifying the formation of code words from code symbols. Compression exceeds entropy by the amount of the informational divergence, between source words and code words, divided by the expected source-word length. The asymptotic optimality of Tunstall or Huffman codes derives from the bounding of divergence while the expected source-word length is made arbitrarily large. A heuristic extension scheme is asymptotically optimal but also acts to reduce the divergence by retaining those source words which are well matched to their corresponding code words.>

Entropy coding introduces a probabilistic matching at the interface between the discrete-time, finite-alphabet, source model sequence and the discrete-time, finite-alphabet, channel information sequence. Source entropy codes, such as common-string parsers, alter the source statistics to match the given channel. Channel entropy codes, such as Huffman codes, alter the channel statistics to match the given source. These are given a joint, dual description using parse and code trees describing the source and channel word sets. Divergence and variational distance are defined between the source word distribution and the desired channel word distribution. Convergence of any scheme for extending the size of the source word set is demonstrated if the divergence divided by the expected source-word length becomes zero. Huffman channel coding implies an upper bound on divergence in terms of variational distance. This leads to an heuristic algorithm to extend the source word set while keeping the divergence low. Examples show a significant complexity reduction for performance equivalent to Tunstall source coding or Huffman channel coding alone.>

Time-invariant trellis codes for stationary, ergodic, discrete-time sources are designed by unconstrained, nonlinear optimization of the performance in a simulated source encoding with the Viterbi algorithm. A nonderivative conjugate directions algorithm and a conjugate gradient algorithm with restarts are applied to design low-constraint-length, unit-rate, binary codes for the memoryless Gaussian source. The latter algorithm is also used to design codes for the memoryless Laplacian source and a third-order autoregressive model for speech. Good codes are tabulated and compared to other known results on a performance versus complexity basis. Those for the Gaussian source are tested in a joint (tandem) trellis-coding system with known convolutional channel codes.>

The design of time-invariant trellis codes for stationary ergodic discrete-time sources is cast as an unconstrained, nonlinear optimization problem, where the objective function and its derivatives are evaluated by simulation. Using classical real analysis and the ergodic theorem, convergence of the sample encoding distortion and its partial derivatives (with respect to the code quantization levels) to their ensemble average values is investigated. It is found that in the common code design situation, the expected per-symbol distortion and its first derivatives are available and piecewise continuous, but second-derivative information is unreliable. This indicates that efficient optimization should be performed using a nonderivative or first-derivative method that does not compute approximate second derivatives to determine search directions.>