scispace - formally typeset
Search or ask a question

Showing papers on "Codebook published in 2002"


Journal ArticleDOI
TL;DR: Experimental results show that successful tool state detection rates as high as 97% can be achieved by using the proposed new modeling framework for tool wear monitoring in machining processes using hidden Markov models.
Abstract: This paper presents a new modeling framework for tool wear monitoring in machining processes using hidden Markov models (HMMs). Feature vectors are extracted from vibration signals measured during turning. A codebook is designed and used for vector quantization to convert the feature vectors into a symbol sequence for the hidden Markov model. A series of experiments are conducted to evaluate the effectiveness of the approach for different lengths of training data and observation sequence. Experimental results show that successful tool state detection rates as high as 97% can be achieved by using this approach.

151 citations


Journal ArticleDOI
TL;DR: An algorithm for automatic image orientation estimation using a Bayesian learning framework is presented and it is demonstrated that a small codebook extracted from a learning vector quantizer can be used to estimate the class-conditional densities of the observed features needed for the Bayesian methodology.
Abstract: We present an algorithm for automatic image orientation estimation using a Bayesian learning framework. We demonstrate that a small codebook (the optimal size of codebook is selected using a modified MDL criterion) extracted from a learning vector quantizer (LVQ) can be used to estimate the class-conditional densities of the observed features needed for the Bayesian methodology. We further show how principal component analysis (PCA) and linear discriminant analysis (LDA) can be used as a feature extraction mechanism to remove redundancies in the high-dimensional feature vectors used for classification. The proposed method is compared with four different commonly used classifiers, namely k-nearest neighbor, support vector machine (SVM), a mixture of Gaussians, and hierarchical discriminating regression (HDR) tree. Experiments on a database of 16 344 images have shown that our proposed algorithm achieves an accuracy of approximately 98% on the training set and over 97% on an independent test set. A slight improvement in classification accuracy is achieved by employing classifier combination techniques.

123 citations


Journal ArticleDOI
TL;DR: The proposed vector quantizer (VQ) in the wavelet domain for the compression of electrocardiogram (ECG) signals outperforms many recently published methods, including the best one known as the set partitioning in hierarchical trees.
Abstract: In this paper, we propose a novel vector quantizer (VQ) in the wavelet domain for the compression of electrocardiogram (ECG) signals. A vector called tree vector (TV) is formed first in a novel structure, where wavelet transformed (WT) coefficients in the vector are arranged in the order of a hierarchical tree. Then, the TVs extracted from various WT subbands are collected in one single codebook. This feature is an advantage over traditional WT-VQ methods, where multiple codebooks are needed and are usually designed separately because numerical ranges of coefficient values in various WT subbands are quite different. Finally, a distortion-constrained codebook replenishment mechanism is incorporated into the VQ, where codevectors can be updated dynamically, to guarantee reliable quality of reconstructed ECG waveforms. With the proposed approach both visual quality and the objective quality in terms of the percent of root-mean-square difference (PRD) are excellent even in a very low bit rate. For the entire 48 records of Lead II ECG data in the MIT/BIH database, an average PRD of 7.3% at 146 b/s is obtained. For the same test data under consideration, the proposed method outperforms many recently published ones, including the best one known as the set partitioning in hierarchical trees.

118 citations


Journal ArticleDOI
TL;DR: A keyblock-based approach to content-based image retrieval where each image is encoded as a set of one-dimensional index codes linked to the keyblocks in the codebook, analogous to considering a text document as a linear list of keywords.
Abstract: The success of text-based retrieval motivates us to investigate analogous techniques which can support the querying and browsing of image data. However, images differ significantly from text both syntactically and semantically in their mode of representing and expressing information. Thus, the generalization of information retrieval from the text domain to the image domain is non-trivial. This paper presents a framework for information retrieval in the image domain which supports content-based querying and browsing of images. A critical first step to establishing such a framework is to construct a codebook of "keywords" for images which is analogous to the dictionary for text documents. We refer to such "keywords" in the image domain as "keyblocks." In this paper, we first present various approaches to generating a codebook containing keyblocks at different resolutions. Then we present a keyblock-based approach to content-based image retrieval. In this approach, each image is encoded as a set of one-dimensional index codes linked to the keyblocks in the codebook, analogous to considering a text document as a linear list of keywords. Generalizing upon text-based information retrieval methods, we then offer various techniques for image-based information retrieval. By comparing the performance of this approach with conventional techniques using color and texture features, we demonstrate the effectiveness of the keyblock-based approach to content-based image retrieval.

97 citations


Journal ArticleDOI
TL;DR: This work alters the encoding of multiple description lattice vector quantization technique to improve performance without a significant increase in complexity by replacing the fine lattice codebook with a nonlattice code book that respects many of the symmetries of the coarse lattice.
Abstract: A multiple description (MD) lattice vector quantization technique for two descriptions was previously introduced in which fine and coarse codebooks are both lattices. The encoding begins with quantization to the nearest point in the fine lattice. This encoding is an inherent optimization for the decoder that receives both descriptions; performance can be improved with little increase in complexity by considering all decoders in the initial encoding step. The altered encoding relies only on the symmetries of the coarse lattice. This allows us to further improve performance without a significant increase in complexity by replacing the fine lattice codebook with a nonlattice codebook that respects many of the symmetries of the coarse lattice. Examples constructed with the two-dimensional (2-D) hexagonal lattice demonstrate large improvement over time sharing between previously known quantizers.

89 citations


Journal ArticleDOI
TL;DR: A wrapped Leech lattice shape quantizer is used to demonstrate a signal-to-quantization-noise ratio within 1 dB of the distortion-rate function for rates above 1 bit per sample, and an improvement over existing techniques of similar complexity.
Abstract: A fixed-rate shape-gain quantizer for the memoryless Gaussian source is proposed. The shape quantizer is constructed from wrapped spherical codes that map a sphere packing in /spl Ropf//sup k-1/ onto a sphere in /spl Ropf//sup k/, and the gain codebook is a globally optimal scalar quantizer. A wrapped Leech lattice shape quantizer is used to demonstrate a signal-to-quantization-noise ratio within 1 dB of the distortion-rate function for rates above 1 bit per sample, and an improvement over existing techniques of similar complexity. An asymptotic analysis of the tradeoff between gain quantization and shape quantization is also given.

88 citations


Patent
Wooyoung Jang1
11 Jan 2002
TL;DR: An apparatus for image coding using tree-structured vector quantization based on a wavelet transform and a method therefor are provided in this paper, where the advantages of wavelet vector quantisation and zerotree coding are maximized so that the size of a codebook is reduced and the coding performance is improved.
Abstract: An apparatus for image coding using tree-structured vector quantization based on a wavelet transform and a method therefor are provided The apparatus for image coding using a tree-structured vector quantization based on wavelet transform has a wavelet transform unit, a vector construct unit, an error vector unit, a scan unit, a first quantization unit and a second quantization unit The wavelet transform unit wavelet transforms an input image signal The vector construct unit constructs vectors, each having a tree structure in a different direction, using the wavelet transformed result The error vector generation unit generates a plurality of error vectors by setting one of the vectors as a basic vector and performing a calculation on each of the vectors remaining with respect to the basic vector The scan unit scans the coefficients of each of the basic vector and error vectors in a different direction The first vector quantization unit generates a first codebook for the basic vector scanned in the scan unit, quantizes the scanned basic vector using the first codebook, and outputs the quantization result as the index of the first codebook The second vector quantization unit generates a second codebook for the error vectors scanned in the scan unit, quantizes the scanned error vectors using the second codebook, and outputs the quantization results as the indices of the second codebook According to the apparatus and method, the advantages of wavelet vector quantization and zerotree-coding are maximized so that the size of a codebook is reduced and the coding performance is improved

60 citations


Patent
17 Jun 2002
TL;DR: A method, apparatus, system, and computer software for compressing and decompressing a message for transmission can be found in this paper, where the method of compressing text messages for transmission may include parsing text strings and encoding numerical values with a binary representation and analyzing values of the text strings.
Abstract: A method, apparatus, system, and computer software for compressing and decompressing a message for transmission. The method of compressing a text message for transmission may include parsing text strings and encoding numerical values with a binary representation and analyzing values of the text strings and populating a session specific codebook with partial strings from the values. The method of compressing a message for transmission may also include parsing the message with a template and generating at least one substring to be transmitted; parsing the at least one substring with entries in a session specific codebook and generating a first part of the compressed message; populating the session specific codebook with entries for unknown field values; parsing any unmatched substrings with entries from a first static dictionary and generating a second part of the compressed message; parsing any still unmatched substrings with entries from a second static dictionary and generating a third part of the compressed message; compressing a remainder of the substrings with a compression algorithm; and combining the first part, the second part, and the third part of the compressed message to obtain a compressed message for transmission.

59 citations


Journal ArticleDOI
TL;DR: This paper extends and modifies classified vector quantization (CVQ) to improve the quality of compressed images and shows that the image quality is improved dramatically.

59 citations


Patent
Takahiro Unno1
27 Feb 2002
TL;DR: In this article, a decoder for code excited LP encoded frames with both adaptive and fixed codebooks was proposed, and erased frame concealment uses repetitive excitation plus a smoothing of pitch gain in the next good frame, plus multilevel voicing classification with multiple thresholds of correlations determining linear interpolated adaptive and codebook excitation contributions.
Abstract: A decoder for code excited LP encoded frames with both adaptive and fixed codebooks; erased frame concealment uses repetitive excitation plus a smoothing of pitch gain in the next good frame, plus multilevel voicing classification with multiple thresholds of correlations determining linear interpolated adaptive and fixed codebook excitation contributions.

55 citations



Journal ArticleDOI
TL;DR: The experimental results show that the proposed mean-removed classified vector quantization (MRCVQ) can remove effectively the artifacts caused by high compression and improve the perceptual quality significantly.

Patent
David E. Penna1
18 Sep 2002
TL;DR: In this paper, variable length decoding of DCT coefficients in MPEG video data is performed using a standard processor ( 400 ) and a small look-up table (LUT 530 ), which performs an integer to floating point conversion on a portion of the received bitstream (BS).
Abstract: Variable length decoding of DCT coefficients in MPEG video data is performed using a standard processor ( 400 ) and a small look-up table (LUT 530 ). The processor performs ( 520 ) an integer to floating point conversion on a portion the received bitstream (BS). By this step, lengthy codewords with many leading zeros, which are common in the codebook, are represented in a compressed form by the exponent and mantissa fields (EXP, MAN) of the floating point result (FP). The relevant bits are extracted and used as an index (IX) to address the LUT. This avoids cumbersome bit-oriented logic, while also avoiding a very large LUT that would otherwise be required to represent the same codebook. The entire LUT may thus reside in cache memory ( 410 ). In a VLIW processor implementation, decoding of one token is pipelined with the inverse scan and inverse quantisation step of the preceding token(s).

Proceedings Article
01 Jan 2002
TL;DR: The experiments show that the use of visual information in AVCDCN allows significant performance gains over CDCN, and the technique is considered for use with both audio-only and audio-visual speech recognition.
Abstract: We introduce a non-linear enhancement technique called audio-visual codebook dependent cepstral normalization (AVCDCN) and we consider its use with both audio-only and audio-visual speech recognition. AVCDCN is inspired from CDCN, an audio-only enhancement technique that approximates the nonlinear effect of noise on speech with a piecewise constant function. Our experiments show that the use of visual information in AVCDCN allows significant performance gains over CDCN.

Proceedings ArticleDOI
04 Aug 2002
TL;DR: In this article, a non-linear enhancement technique called audio-visual codebook dependent cepstral normalization (AVCDCN) was proposed for both audio-only and audio-Visual speech recognition.
Abstract: We introduce a non-linear enhancement technique called audio-visual codebook dependent cepstral normalization (AVCDCN) and we consider its use with both audio-only and audio-visual speech recognition. AVCDCN is inspired from CDCN, an audio-only enhancement technique that approximates the nonlinear effect of noise on speech with a piecewise constant function. Our experiments show that the use of visual information in AVCDCN allows significant performance gains over CDCN.

Journal ArticleDOI
01 Aug 2002
TL;DR: The existence of universal mixture codebooks are demonstrated, and it is shown that it is possible to universally encode memoryless sources with redundancy of approximately (d/2) log n bits, where d is the dimension of the simplex of probability distributions on the reproduction alphabet.
Abstract: We characterize the best achievable performance of lossy compression algorithms operating on arbitrary random sources, and with respect to general distortion measures. Direct and converse coding theorems are given for variable-rate codes operating at a fixed distortion level, emphasizing: (a) nonasymptotic results, (b) optimal or near-optimal redundancy bounds, and (c) results with probability one. This development is based in part on the observation that there is a precise correspondence between compression algorithms and probability measures on the reproduction alphabet. This is analogous to the Kraft inequality in lossless data compression. In the case of stationary ergodic sources our results reduce to the classical coding theorems. As an application of these general results, we examine the performance of codes based on mixture codebooks for discrete memoryless sources. A mixture codebook (or Bayesian codebook) is a random codebook generated from a mixture over some class of reproduction distributions. We demonstrate the existence of universal mixture codebooks, and show that it is possible to universally encode memoryless sources with redundancy of approximately (d/2) log n bits, where d is the dimension of the simplex of probability distributions on the reproduction alphabet.

Patent
Dae-Ryong Lee1
23 May 2002
TL;DR: In this paper, a fixed codebook searcher was proposed for searching an excitation (or fixed) codebook in a speech coding system including a synthesis filter for synthesizing a speech signal.
Abstract: A method for searching an excitation (or fixed) codebook in a speech coding system. In a speech coding system including a synthesis filter for synthesizing a speech signal, a fixed codebook searcher according to the present invention segments a speech signal frame into a plurality of subframes to generate an excitation signal to be used in a synthesis filter, segments again each of the subframes into a plurality of subgroups, and searches the respective subframes each comprised of a plurality of pulse position/amplitude combinations for pulses. The fixed codebook searcher searches the respective subgroups for a predetermine number of pulses having non-zero amplitude, and generates the searched pulses as an initial vector. Next, the fixed codebook searcher selects a pulse combination including at least one pulse among the pulses of the initial vector, and then substitutes pulses of the selected pulse combination for pulses in other positions in the subgroups. The selection and the substitution are repeatedly performed on all the pulses of the initial vector.

Journal ArticleDOI
TL;DR: A fast Kohonen self-organizing neural network algorithm is used to achieve big saving in codebook construction time and propose a new approach called the transformed vector quantization (TVQ), combining the features of transform coding and VQ.

Proceedings ArticleDOI
06 Oct 2002
TL;DR: A method for coding "unconstrained" fixed codebook (FCB) excitation for ACELP speech coders and it is proposed that the unconstrained FCB does not place track-based constraint on the pulse positions.
Abstract: A method for coding "unconstrained" fixed codebook (FCB) excitation for ACELP speech coders is proposed. The unconstrained FCB does not place track-based constraint on the pulse positions. The coding method combines Huffman codes and combinatorial codes. The method is less sensitive to bit errors and is nearly as efficient as the combinatorial codes. A method for efficiently storing the parameters needed in the combinatorial codebook is also proposed.

Patent
31 Oct 2002
TL;DR: In this article, a document page is encoded into codes associated with various locations of the document page, and the codes are assembled into a code book, and then the captured images are similarly encoded and searched against the codes in the codebook.
Abstract: A system and method for encoding a document image and finding a location based on that image are described. A document page is encoded into codes associated with various locations of the document page. The codes are assembled into a code book. Captured images may then be similarly encoded and searched against the codes in the codebook. One or more codes and associated locations may be returned, thereby providing one or more possible locations for the captured images.

Proceedings ArticleDOI
07 Jan 2002
TL;DR: This paper proposes the comparison of the codebooks, based on a Modified Hausdorff distance, as a novel method for compressed domain image retrieval, and shows this technique to give excellent results, outperforming classical color indexing techniques.
Abstract: Image retrieval and image compression are both very active fields of research. Unfortunately, in the past they were pursued independently leading to image indexing methods being both efficient and effective but restricted to uncompressed images. In this paper we introduce an image retrieval technique that operates in the compressed domain of vector quantize images. Vector quantization (VQ) achieves compression by representing image blocks as indices into a codebook of prototype blocks. By realizing that, if images are coded with their own VQ codebook then much of the image information is contained in the codebook itself, we propose the comparison of the codebooks, based on a Modified Hausdorff distance, as a novel method for compressed domain image retrieval. Experiments, based on an image database comprising many colorful pictures show this technique to give excellent results, outperforming classical color indexing techniques.

Proceedings ArticleDOI
01 Dec 2002
TL;DR: A codebook design method is based on a quantized version of traditional maximum ratio transmission with maximum ratio combining at the receiver, which exploits the quantization problem's relationship with Grassmannian line packaging.
Abstract: Multiple-input multiple-output wireless systems can provide substantial gains in capacity and quality compared to single-input single-output (SISO) wireless systems. Maximum ratio transmission has been shown to be a low complexity solution to improving average signal-to-noise ratio (SNR), however, it requires feedback. Since in practice full channel knowledge at the transmitter is difficult to realize, we propose a technique where the receiver sends channel state information in the form of a codebook label. A codebook design method is based on a quantized version of traditional maximum ratio transmission with maximum ratio combining at the receiver. The codebook design criterion exploits the quantization problem's relationship with Grassmannian line packaging. Systems using the transmit diversity codebooks are shown to have a diversity order of the product of the number of transmit and the number of receive antennas. Monte Carlo simulations compare the performance of systems using this new codebook method with previously proposed systems.

Book ChapterDOI
TL;DR: A weighted distortion measure is introduced that takes into account the correlations between the known models in the speaker database and larger weights are assigned to vectors that have high discriminating power between the speakers and vice versa.
Abstract: We consider the distortion measure in vector quantization based speaker identification system. The model of a speaker is a codebook generated from the set of feature vectors from the speakers voice sample. The matching is performed by evaluating the distortions between the unknown speech sample and the models in the speaker database. In this paper, we introduce a weighted distortion measure that takes into account the correlations between the known models in the database. Larger weights are assigned to vectors that have high discriminating power between the speakers and vice versa.

Proceedings ArticleDOI
06 Oct 2002
TL;DR: A unified framework for the efficient ACELP codebook search method based on the pulse replacement procedure is proposed, and it is confirmed that the proposed method provides the same performance as the standard method with much less search load.
Abstract: Speech codecs with ACELP structure are widely used due to their high performance for the bit rates between 4 kbps and 16 kbps. In ACELP codecs, the search for the best code vector from a huge codebook is the crucial part in both performance and complexity. In most cases, a non-optimal search method is used in order to reduce the search load with minimal performance degradation. However, its complexity is still the major part of the overall codec complexity. Furthermore, there is no unified method for ACELP codebook search; a rather different method is developed for each ACELP codec. In this paper, a unified framework for the efficient ACELP codebook search method based on the pulse replacement procedure is proposed, and is applied to standard ACELP codecs. It is confirmed that the proposed method provides the same performance as the standard method with much less search load.

Patent
29 Apr 2002
TL;DR: In this article, an apparatus and method for preventing a computation error of band-by-band bit allocation between an encoder and a decoder is presented. But the method is not suitable for the case where the encoder uses a quantization device to quantize an input LSP (line spectrum pair) coefficient, and a quantized output is output.
Abstract: An apparatus and method for preventing a computation error of band-by-band bit allocation between an encoder and a decoder. In the encoder, an input LSP (line spectrum pair) coefficient is quantized by a quantization device, and a quantized output is output. In a codebook referring section, the look up of a codebook in which computation results are prestored is performed by using an LSP index of the first stage, and band-by-band bit allocation information is created. LSP indexes for each quantization stage are supplied to a decoder. In the decoder, the look up of the codebook is performed using the LSP index, and dequantization is performed on the basis of the created band-by-band bit allocation information.

01 Jul 2002
TL;DR: The Splitting solution to implement the Codebook, which improves the image quality by the average Training Vectors, then splits the average result to Codebook that has minimum distortion, will give the better quality of the image than using Random Codebook.
Abstract: Compressing image data by using Vector Quantization (VQ)[1]-[3] will compare Training Vectors with Codebook. The result is an index of position with minimum distortion. The implementing Random Codebook will reduce the image quality. This research presents the Splitting solution [4],[5] to implement the Codebook, which improves the image quality[6] by the average Training Vectors, then splits the average result to Codebook that has minimum distortion. The result from this presentation will give the better quality of the image than using Random Codebook.

Proceedings ArticleDOI
07 Aug 2002
TL;DR: A 3-step hierarchical fast search algorithm is proposed by narrowing search scope, skipping redundant distance computation and lastly simplifying must-do distance computation to encode an image using VQ fast.
Abstract: In order to encode an image using VQ fast, it is most important to eliminate unnecessary distance computation as much as possible while searching for winner code. Sum and partial sum are used as features in this paper to roughly measure the difference between an input image block and a code to check whether current code could be a promising candidate winner code or not. A 3-step hierarchical fast search algorithm is proposed by narrowing search scope, skipping redundant distance computation and lastly simplifying must-do distance computation. For 10 standard gray-level images of size 512/spl times/512 with very different details, computational complexity can be reduced to below 5% ultimately for a codebook of size 1024 or 2048 meanwhile keeping the PSNR not as degraded as that of full search.

Patent
Anssi Rämö1
10 May 2002
TL;DR: In this paper, a method and system for quantizing LSF vectors in a speech coder is presented, where predicted LSF values based on previously decoded output values are used to estimate spectral distortion, along with the residual codebook vectors and the LSF coefficients.
Abstract: A method and system for quantizing LSF vectors in a speech coder, wherein predicted LSF values based on previously decoded output values are used to estimate spectral distortion, along with the residual codebook vectors and the LSF coefficients. The method comprises the steps of obtaining a plurality of quantized LSF coefficients from the respective predicted LSF values (28), and the residual codebook vectors; rearranging the quantized LSF coefficients in the frequency domain in an orderly fashion (40); obtaining the spectral distortion from the rearranged quantized LSF coefficients and the respective LSF coefficients (42); and an optimal code vector is selected based on the spectral distortion (44).

Patent
08 Jan 2002
TL;DR: In this paper, a synthesis filter is implemented using linear prediction coefficients obtained by dividing an input signal into frames each of a fixed length, and subjecting the input signal to linear prediction analysis in the frame units.
Abstract: Disclosed is a voice encoding method having a synthesis filter implemented using linear prediction coefficients obtained by dividing an input signal into frames each of a fixed length, and subjecting the input signal to linear prediction analysis in the frame units, generating a reconstructed signal by driving said synthesis filter by a periodicity signal output from an adaptive codebook and a pulsed signal output from an algebraic codebook, and performing encoding in such a manner that an error between the input signal and said reproduced signal is minimized, wherein there are provided an encoding mode 1 that uses pitch lag obtained from an input signal of a present frame and an encoding mode 2 that uses pitch lag obtained from an input signal of a past frame. Encoding is performed in encoding mode 1 and encoding mode 2, the mode in which the input signal can be encoded more precisely is decided frame by frame and encoding is carried out on the basis of the mode decided.

Journal ArticleDOI
TL;DR: Simulation results show that 15%-20% of the bit rates in the noniterative FBC technique are saved by using the proposed GMFVQs, which are two classes of finite state fractal vector quantizers for the image coding framework.
Abstract: In this paper, we propose gradient match fractal vector quantizers (GMFVQs) and side match fractal vector quantizers (SMFVQs), which are two classes of finite state fractal vector quantizers (FSFVQs), for the image coding framework. In our previous work, we proposed the noniterative fractal block coding (FBC) technique to improve the decoding speed and the coding performance for conventional FBC techniques. To reduce the number of bits for denoting the fractal code of the range block, the concepts of the gradient match vector quantizers (GMVQs) and the side match vector quantizers (SMVQs) are employed to the noniterative FBC technique. Unlike ordinary vector quantizers, the super codebooks in the proposed GMFVQs and SMFVQs are generated from the affine-transformed domain blocks in the noniterative FBC technique. The codewords in the state codebook are dynamically extracted from the super codebook with the side-match and gradient-match criteria. The redundancy in the affine-transformed domain blocks is greatly reduced and the compression ratio can be significantly increased. Our simulation results show that 15%-20% of the bit rates in the noniterative FBC technique are saved by using the proposed GMFVQs.