Vector Quantization Method Research Articles

Image Compression Using Vector Quantization and Genetic Algorithms

Image compression involves reducing the size of image data file, while retaining necessary information.This paper uses the facilities of the Genetic Algorithm for the enhancement of the performance of one of the popular compression method, Vector Quantization method is selected in this work. After studying this method, new proposed algorithm for mixing the Genetic Algorithm with this method was constructed and then the required programs for testing this algorithm was written. The proposed algorithm was tested by applying it on some image data files. Some fidelity measures are calculated to evaluate the performance of the new proposed algorithm. A good enhancement was recorded for the performance of the Vector Quantization method when mixed with the Genetic Algorithm. All programs were written by using Matlab (version 7.0) and these programs were executed on the Pentium III (866 MHz) personal computer.

Performance evaluation of a novel sampling-based texture synthesis technique using different sized patches

We propose a novel sampling-based texture synthesis algorithm called Multipatch, which improves on the results of previous sampling-based algorithms by using patches of different size, and by minimizing global pasting errors. A key feature of the proposed algorithm is that it always converges to a local minimum. Multipatch, the patchwork algorithm, and Wei and Levoy’s multi-resolution texture synthesis algorithm, which is based on a tree-structured vector quantization method, are statistically analyzed and subjectively evaluated. The results of simulations show that the patchwork algorithm yields a perceptually acceptable texture in a shorter expected running time than the other two algorithms; however, Multipatch is the most efficient in terms of obtaining a good quality texture image.

Optimum switched split vector quantization of LSF parameters

We address the issue of rate–distortion (R/D) performance optimality of the recently proposed switched split vector quantization (SSVQ) method. The distribution of the source is modeled using Gaussian mixture density and thus, the non-parametric SSVQ is analyzed in a parametric model based framework for achieving optimum R/D performance. Using high rate quantization theory, we derive the optimum bit allocation formulae for the intra-cluster split vector quantizer (SVQ) and the inter-cluster switching. For the wide-band speech line spectrum frequency (LSF) parameter quantization, it is shown that the Gaussian mixture model (GMM) based optimum parametric SSVQ method provides 1 bit/vector advantage over the non-parametric SSVQ method.

Switched Conditional PDF-Based Split VQ Using Gaussian Mixture Model

In this letter, we develop switched conditional PDF-based split vector quantization (SCSVQ) method using the recently proposed conditional PDF-based split vector quantizer (CSVQ). The use of CSVQ allows us to alleviate the coding loss by exploiting the correlation between subvectors, in each switching region. Using the Gaussian mixture model (GMM)-based parametric framework, we also address the rate-distortion (R/D) performance optimality of the proposed SCSVQ method by allocating the bits optimally among the switching regions. For the wideband speech line spectrum frequency (LSF) parameter quantization, it is shown that the optimum parametric SCSVQ method provides nearly 2 bits/vector advantage over the recently proposed nonparametric switched split vector quantization (SSVQ) method.

IMAGE: A New Tool for the Prediction of Transcription Factor Binding Sites

IMAGE is an application tool, based on the vector quantization method, aiding the discovery of nucleotidic sequences corresponding to Transcription Factor binding sites. Starting from the knowledge of regulation regions of a number of co-expressed genes, the software is able to predict the occurrence of specific motifs of different lengths (starting from 6 base pairs) with a defined number of punctual mutations.

Improvement of Robustness of Odor Classification in Dynamically Changing Concentration Against Environmental Change

In this paper, we propose a method for improving the robustness of odor classification against humidity change when the odor concentration changes dynamically. We apply a short-time Fourier transform (STFT) to sensor responses to obtain the frequency characteristics, and then employ a stepwise discriminant analysis to select the frequency components effective for the odor classification. We improve the classification performance by selecting the components robust against humidity change and combining them with humidity data. Using a learning vector quantization (LVQ) method, we successfully achieved high classification rate even if the odor concentration changed dynamically and irregularly at different humidity levels whereas the classification rate was insufficient in the case of using only magnitudes of sensor responses.

Optimum Transform Domain Split VQ

In this letter, we develop an optimum transform domain split vector quantization (TrSVQ) method. We address both the issues of achieving best rate-distortion (R/D) performance and less complexity. For quantizing a multivariate Gaussian source, we derive the mean-square error (MSE) performance expression for the TrSVQ method using high rate theory and optimum bit allocation. Also, to reduce the complexity, we develop a binary split- based iterative algorithm and use the algorithm in a tree structured manner to find the optimum subvectors' dimensions (i.e., optimum splits).

MIMO Transmit Beamforming Under Uniform Elemental Power Constraint

We consider multi-input multi-output (MIMO) transmit beamforming under the uniform elemental power constraint. This is a nonconvex optimization problem, and it is usually difficult to find the optimal transmit beamformer. First, we show that for the multi-input single-output (MISO) case, the optimal solution has a closed-form expression. Then we propose a cyclic algorithm for the MIMO case which uses the closed-form MISO optimal solution iteratively. The cyclic algorithm has a low computational complexity and is locally convergent under mild conditions. Moreover, we consider finite-rate feedback methods needed for transmit beamforming. We propose a simple scalar quantization method, as well as a novel vector quantization method. For the latter method, the codebook is constructed under the uniform elemental power constraint and the method is referred as VQ-UEP. We analyze VQ-UEP performance for the MISO case. Specifically, we obtain an approximate expression for the average degradation of the receive signal-to-noise ratio (SNR) caused by VQ-UEP. Numerical examples are provided to demonstrate the effectiveness of our proposed transmit beamformer designs and the finite-rate feedback techniques.

Analysis of Conditional PDF-Based Split VQ

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> The split vector quantization (SVQ) method results in a “coding loss” due to the independent quantization of split subvectors. To recover the coding loss, conditional pdf-based split vector quantization methods were proposed recently. </para>

Classification of mass-spectrometric data in clinical proteomics using learning vector quantization methods

In the present contribution we propose two recently developed classification algorithms for the analysis of mass-spectrometric data-the supervised neural gas and the fuzzy-labeled self-organizing map. The algorithms are inherently regularizing, which is recommended, for these spectral data because of its high dimensionality and the sparseness for specific problems. The algorithms are both prototype-based such that the principle of characteristic representants is realized. This leads to an easy interpretation of the generated classifcation model. Further, the fuzzy-labeled self-organizing map is able to process uncertainty in data, and classification results can be obtained as fuzzy decisions. Moreover, this fuzzy classification together with the property of topographic mapping offers the possibility of class similarity detection, which can be used for class visualization. We demonstrate the power of both methods for two exemplary examples: the classification of bacteria (listeria types) and neoplastic and non-neoplastic cell populations in breast cancer tissue sections.

Vector quantization of mel‐cepstral coefficients using distortion measure for spectral analysis

AbstractThis paper proposes a method for vector quantization of mel‐cepstral coefficients based on a measure for spectral analysis. It is a method in which the match between the input signal within the frame and the mel‐cepstral coefficients stored in the codebook is evaluated by using a statistical measure for spectral analysis, and the quantized mel‐cepstral coefficients are derived. The codebook training algorithm for the proposed method is also presented. The proposed method is further applied to multistage vector quantization. It is shown by an experiment using a speech database that the proposed method with a codebook size of 9 bits can achieve the same quantization distortion as the conventional 12‐bit method (vector quantization based on Euclidean distance between mel‐cepstral coefficients). © 2007 Wiley Periodicals, Inc. Syst Comp Jpn, 38(13): 104–114, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.10432

Conditional PDF-Based Split Vector Quantization of Wideband LSF Parameters

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> The commonly used split vector quantization (SVQ) method is inferior to unconstrained quantization due to independent coding of the split subvectors, resulting in a coding loss. In this paper, we propose a conditional pdf-based split vector quantization (CSVQ) method to recover the coding loss. The CSVQ method is developed assuming the line spectrum frequency source distribution as a multivariate Gaussian and the subvectors are quantized sequentially to exploit the correlation between the subvectors. The new CSVQ method is evaluated for wideband speech LSF quantization; CSVQ is shown to outperform traditional SVQ and provide comparable performance to the recently proposed switched split vector quantization (SSVQ) method. In addition, the transform domain SVQ method is also realized to show that its performance is limited by the distance measure used in the transform domain. </para>

Knowledge-Dependent Pattern Classification of Human Nasal Profiles

(1) To determine feature vector representations (geometric pattern parameters) that are effective in describing human nasal profiles, (2) to determine the number of code vectors (typical nasal patterns) that are mathematically optimized by applying the vector quantization method to each feature vector extracted for each subject, and (3) to determine the morphological traits of each code. Lateral facial photographs of 200 Japanese women recorded for orthodontic diagnosis were selected. Five anatomic landmarks were identified on each image together with a set of data points that constituted the contour of the facial profile. An eight-dimensional feature vector effective in distinguishing differences in nasal profile patterns was extracted from the data set using experts' knowledge of the anatomic traits of the nose. The vector quantization technique was applied to the feature vectors to provide the optimum number of nasal profile patterns. The number of code vectors mathematically optimized was six, and the differences between vectors were maximized by morphological traits of the root, dorsum, tip, and base of the nose. Proportions of the number of image records classified into each code were 25.5%, 24.5%, 21.5%, 15.0%, 10.0%, and 3.5% from code 1 to code 6, respectively. Classifying nasal profile patterns based on knowledge from a linguistic description was found to be more effective than a method based on uniform sectioning. The differences between vectors were maximized by morphological traits of the root, the dorsum, the tip, and the base of the nose.

1D BAR CODE READING ON CAMERA PHONES

The availability of camera phones provides people with a mobile platform for decoding bar codes, whereas conventional scanners lack mobility. However, using a normal camera phone in such applications is challenging due to the out-of-focus problem. In this paper, we present the research effort on the bar code reading algorithms using a VGA camera phone, NOKIA 7650. EAN-13, a widely used 1D bar code standard, is taken as an example to show the efficiency of the method. A wavelet-based bar code region location and knowledge-based bar code segmentation scheme is applied to extract bar code characters from poor-quality images. All the segmented bar code characters are input to the recognition engine, and based on the recognition distance, the bar code character string with the smallest total distance is output as the final recognition result of the bar code. In order to train an efficient recognition engine, the modified Generalized Learning Vector Quantization (GLVQ) method is designed for optimizing a feature extraction matrix and the class reference vectors. 19 584 samples segmented from more than 1000 bar code images captured by NOKIA 7650 are involved in the training process. Testing on 292 bar code images taken by the same phone, the correct recognition rate of the entire bar code set reaches 85.62%. We are confident that auto focus or macro modes on camera phones will bring the presented method into real world mobile use.

Improvement of capability for classifying odors in dynamically changing concentration using QCM sensor array and short-time Fourier transform

In this paper, we propose a method for improving the capability of odor classification in dynamical change of concentration often encountered in the ambient air. Quartz crystal microbalance (QCM) sensors are used to measure the sensor responses in this research. Our proposed method employs a short-time Fourier transform (STFT) algorithm and a stepwise discriminant analysis for feature extraction and dimensional reduction. The STFT is not only simple and easy to understand, but gives frequency characteristics with temporal information required for a real-time application. The variable selection due to the stepwise method also reduces the dimensions of pattern vectors that lead stable classification results. Finally, using a learning vector quantization (LVQ) method to evaluate the classification performance, we successfully achieved a high classification rate even if the odor concentration changed into various conditions whereas, the classification rate was insufficient in the case of using only magnitudes of sensor responses. This shows the robustness of the method against dynamical change of odor concentration.

Forecasting the CATS benchmark with the Double Vector Quantization method

The Double Vector Quantization (DVQ) method, a long-term forecasting method based on the self-organizing maps algorithm, has been used to predict the 100 missing values of the CATS competition data set. An analysis of the proposed time series is provided to estimate the dimension of the auto-regressive part of this nonlinear auto-regressive forecasting method. Based on this analysis experimental results using the DVQ method are presented and discussed. As one of the features of the DVQ method is its ability to predict scalars as well as vectors of values, the number of iterative predictions needed to reach the prediction horizon is further observed. The method stability for the long term allows obtaining reliable values for a rather long-term forecasting horizon.

Effective Image Retrieval Based on Hidden Concept Discovery in Image Database

This paper addresses content-based image retrieval in general, and in particular, focuses on developing a hidden semantic concept discovery methodology to address effective semantics-intensive image retrieval. In our approach, each image in the database is segmented into regions associated with homogenous color, texture, and shape features. By exploiting regional statistical information in each image and employing a vector quantization method, a uniform and sparse region-based representation is achieved. With this representation, a probabilistic model based on statistical-hidden-class assumptions of the image database is obtained, to which the expectation-maximization technique is applied to analyze semantic concepts hidden in the database. An elaborated retrieval algorithm is designed to support the probabilistic model. The semantic similarity is measured through integrating the posterior probabilities of the transformed query image, as well as a constructed negative example, to the discovered semantic concepts. The proposed approach has a solid statistical foundation; the experimental evaluations on a database of 10000 general-purposed images demonstrate its promise and effectiveness.

Adaptive scene dependent filters for segmentation and online learning of visual objects

We propose the adaptive scene dependent filter (ASDF) hierarchy for unsupervised learning of image segmentation, which integrates several processing pathways into a flexible, highly dynamic, and real-time capable vision architecture. It is based on forming a combined feature space from basic feature maps like, color, disparity, and pixel position. To guarantee real-time performance, we apply an enhanced vector quantization method to partition this feature space. The learned codebook defines corresponding best-match segments for each prototype and yields an over-segmentation of the object and the surround. The segments are recombined into a final object segmentation mask based on a relevance map, which encodes a coarse bottom-up hypothesis where the object is located in the image. We apply the ASDF hierarchy for preprocessing input images in a feature-based biologically motivated object recognition learning architecture and show experiments with this real-time vision system running at 6 Hz including the online learning of the segmentation. Because interaction with user is not perfect, the real-world system acquires useful views effectively only at about 1.5 Hz, but we show that for training a new object one hundred views taking only one minute of interaction time is sufficient.

O®-Line Chinese Signature Veri¯cation Based on Segmentation and HMM

Automatic off-line Chinese signature verification is a very complicated problem. The difficulty lies in the fact that it is hard to find a signature model that is insusceptible to intro-classes distance and at the same time is sensitive to inter-classes distance. In this paper, a simple robust segmentation method with low computation cost is proposed which can successfully extract strokes of handwritten Chinese characters and takes into account the characteristics of signature verification. After being segmented and feature extracted, each signature is represented by a series of six-dimensional vectors quantized using an improved vector quantization method to obtain a series of observation values. Twelve genuine samples were used to train the signature DHMM of a writer and 4576 signatures are used in the test. The cross error rate is only 5.5%.

Self-organizing maps with dynamic learning for signal reconstruction

Wireless Brain Machine Interface (BMI) communication protocols are faced with the challenge of transmitting the activity of hundreds of neurons which requires large bandwidth. Previously a data compression scheme for neural activity was introduced based on Self Organizing Maps (SOM). In this paper we propose a dynamic learning rule for improved training of the SOM on signals with sparse events which allows for more representative prototype vectors to be found, and consequently better signal reconstruction. This work was developed with BMI applications in mind and therefore our examples are geared towards this type of signals. The simulation results show that the proposed strategy outperforms conventional vector quantization methods for spike reconstruction.