Trellis-coded quantization (TCQ) is incorporated into a transform coding structure for encoding monochrome images. Both fixed-rate and entropy-constrained designs are considered. The fixed-rate TCQ-based systems provi...
详细信息
Trellis-coded quantization (TCQ) is incorporated into a transform coding structure for encoding monochrome images. Both fixed-rate and entropy-constrained designs are considered. The fixed-rate TCQ-based systems provide gains in peak signal-to-noise ratio of up to 4.13 dB over scalar quantizer-based schemes, while the entropy-constrained TCQ-based systems provide gains of up to 6.39 dB. The subjective quality of the TCQ-based systems is substantially better than that of comparable scalar quantizer-based systems. The high-frequency background noise, fuzziness, and block artifacts produced by the scalar quantizer-based systems are virtually eliminated by the TCQ-based systems.< >
The coding of high-quality sound at 64 kb/s is of interest for applications such as ISDN. The algorithm described allows the reduction to such a bit rate while maintaining the original quality. It is based on transfor...
详细信息
The coding of high-quality sound at 64 kb/s is of interest for applications such as ISDN. The algorithm described allows the reduction to such a bit rate while maintaining the original quality. It is based on transform coding, and uses a time-domain aliasing cancellation (TDAC) transformation. Perceptual properties and the interblock redundancy of the spectrum are involved when coding the transform coefficients. The complexity of the algorithm allows its real-time implementation on a one floating-point digital signal processor, such as the ATT DSP 32C. The performance and subjective results of the coding system are discussed.< >
We examine the problem of large scale nearest neighbor search in high dimensional spaces and propose a new approach based on the close relationship between nearest neighbor search and that of signal representation and...
详细信息
We examine the problem of large scale nearest neighbor search in high dimensional spaces and propose a new approach based on the close relationship between nearest neighbor search and that of signal representation and quantization. Our contribution is a very simple and efficient quantization technique using transform coding and product quantization. We demonstrate its effectiveness in several settings, including large-scale retrieval, nearest neighbor classification, feature matching, and similarity search based on the bag-of-words representation. Through experiments on standard data sets we show it is competitive with state-of-the-art methods, with greater speed, simplicity, and generality. The resulting compact representation can be the basis for more elaborate hierarchical search structures for sub-linear approximate search. However, we demonstrate that optimized linear search using the quantized representation is extremely fast and trivially parallelizable on modern computer architectures, with further acceleration possible by way of GPU implementation.
Accelerometer data collected from moving vehicles can be modeled as a self-similar random process. A Renyi entropy measure computed over the Wigner-Ville distribution of this non-stationary process is used to select t...
详细信息
ISBN:
(纸本)9781479934188
Accelerometer data collected from moving vehicles can be modeled as a self-similar random process. A Renyi entropy measure computed over the Wigner-Ville distribution of this non-stationary process is used to select the most relevant data samples for compression. Wavelet transform based transform coding is applied to compress the accelerometer data with minimal distortion and accurate inferences (detection of events) on the reconstructed data using the Cher off distance.
An investigation is presented of transform-based seismic data compression. The study concentrates on discrete orthogonal transforms such as the discrete Fourier transform (DFT), the discrete cosine transform (DCT), th...
详细信息
An investigation is presented of transform-based seismic data compression. The study concentrates on discrete orthogonal transforms such as the discrete Fourier transform (DFT), the discrete cosine transform (DCT), the Walsh-Hadamard transform (WHT), and the Karhunen-Loeve transform (KLT). Uniform and subband transform coding schemes were implemented, and comparative results are given for data rates ranging from 150 to 550 b/s. These results are also compared to existing linear prediction techniques.< >
An image coding technique is described that compresses monochrome digital TV images from 8 to about 1 bit/pixel while maintaining high quality. First, for subblocks of 8*8 pixels the discrete cosine transform is calcu...
详细信息
An image coding technique is described that compresses monochrome digital TV images from 8 to about 1 bit/pixel while maintaining high quality. First, for subblocks of 8*8 pixels the discrete cosine transform is calculated. Then, the resulting block of 8*8 transform coefficients is divided into a number of subvectors, each of which is normalized and quantized using vector quantization. The subvector construction and the vector quantization are performed adaptively to the 'direction' of spatial activity in the pixel subblock. The quantization also adapts to the energy of the subvector.< >
The pyramid vector quantizer (PVQ) is a lattice quantizer that was motivated bv the geometric properties of a memoryless Laplacian source. For large rates and the cubic lattice. the PVQ provides improvements of 2.39, ...
详细信息
The pyramid vector quantizer (PVQ) is a lattice quantizer that was motivated bv the geometric properties of a memoryless Laplacian source. For large rates and the cubic lattice. the PVQ provides improvements of 2.39, 5.64, and 8.40 dB for memoryless Gaussian, Laplacian, and gamma sources, respectivelv, compared to the corresponding optimum (noniuniform) scalar quantizer. The lattice basis of the PVQ allows simple encoding and decoding algorithms with a complexity that grows only linearly with the vector dimension. A correlated source such as speech has a geometric nature that is not well suited to the PVQ unless transform coding is used. It is demonstrated that an encoding system using a cosine transform, interleaving of the transform coefficients, and pyramid vector quantization can achieve signal-to-noise ratio (SNR) performance in excess of 20dB for an average rate of 2 bits/sample.
This paper considers the problem of universal transform coding based on estimating the Karhunen-Loeve transform from quantized data. The use of quantized data in the estimation allows the encoder and decoder to mainta...
详细信息
This paper considers the problem of universal transform coding based on estimating the Karhunen-Loeve transform from quantized data. The use of quantized data in the estimation allows the encoder and decoder to maintain the same state without any side information. A theorem is presented that proves, under certain conditions, that consistent estimation of all the required moments is possible from uniformly scalar quantized data regardless of the quantization coarseness. This consistent estimation requires the solution of nonlinear equations. Very simple approximations that avoid these nonlinear equations are used to develop a practical adaptive coding technique. Promising experimental results obtained with this method are presented.
With mean-squared error D as a goal, it is well known that one may approach the rate-distortion function R(D) of a nonbandlimited, continuous-time Gaussian source by sampling at a sufficiently high rate, applying the ...
详细信息
ISBN:
(纸本)9781424413973;1424413974
With mean-squared error D as a goal, it is well known that one may approach the rate-distortion function R(D) of a nonbandlimited, continuous-time Gaussian source by sampling at a sufficiently high rate, applying the Karhunen-Loeve transform to sufficiently long blocks, and then independently coding the transform coefficients of each type. In particular, the coefficients of a given type are ideally encoded with performance attaining a suitably chosen point on the first-order rate-distortion function of that type of coefficient. This paper considers a similar sample-and-transform coding scheme in which ideal coding of coefficients is replaced by coding with some specified family of quantizers, whose operational rate-distortion function is convex. A prime example is scalar quantization with entropy-coding and, if needed for convexity, time sharing. It is shown that when the sampling rate is large, the operational rate-distortion function of such a scheme comes within a finite constant of R(D). Applied to the scalar quantization family, the finiteness of this bound contrasts with a recent result showing that direct scalar quantization of samples (without a transform) has unbounded rate when distortion is held constant and sampling rate becomes large, even when the quantized samples are compressed to their entropy-rate. Thus, at high sampling rates, the transform reduces the loss due to scalar quantization from something infinite to something finite.
暂无评论