Vector quantization in speech coding

Makhoul, John; Roucos, S.; Gish, H.

doi:10.1109/proc.1985.13340

Cited by 731 publications

(226 citation statements)

References 107 publications

Supporting

Mentioning

211

Contrasting

Unclassified

Order By: Relevance

“…The TRN algorithm Martinetz and Schulten, 1994) offers a flexible way to develop a discrete representation of biological data including neighborhood relationships . It can be shown that the algorithm minimizes the so-called distortion error (Makhoul et al, 1985;Martinetz and Schulten, 1994) of the discrete representation. From a signal processing point of view this error measures the fidelity of the data encoding, i.e., the meansquares deviation of the discrete set of vectors from the corresponding data (Makhoul et al, 1985).…”

Section: Design Of the Situs Packagementioning

confidence: 99%

“…It can be shown that the algorithm minimizes the so-called distortion error (Makhoul et al, 1985;Martinetz and Schulten, 1994) of the discrete representation. From a signal processing point of view this error measures the fidelity of the data encoding, i.e., the meansquares deviation of the discrete set of vectors from the corresponding data (Makhoul et al, 1985). If the global minimum of the error is found, the vector distribution is fully determined by the input density distribution.…”

Section: Design Of the Situs Packagementioning

confidence: 99%

“…Key to the docking procedure are topology-representing neural networks (TRNs) that are used to correlate the high-and low-resolution data sets . TRNs are one of several possible algorithmic realizations of vector quantization, a data compression technique frequently used in image and speech processing applications (Gray, 1984;Makhoul et al, 1985). The representation of data by a small number of so-called codebook vectors reduces the combinatorial complexity of the structural comparison, as data sets are compared indirectly by the corresponding vectors.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Situs: A Package for Docking Crystal Structures into Low-Resolution Maps from Electron Microscopy

Wriggers

Milligan

McCammon

1999

Journal of Structural Biology

499

492

View full text Add to dashboard Cite

Three-dimensional image reconstructions of largescale protein aggregates are routinely determined by electron microscopy (EM). We combine lowresolution EM data with high-resolution structures of proteins determined by x-ray crystallography. A set of visualization and analysis procedures, termed the Situs package, has been developed to provide an efficient and robust method for the localization of protein subunits in low-resolution data. Topologyrepresenting neural networks are employed to vector-quantize and to correlate features within the structural data sets. Microtubules decorated with kinesin-related ncd motors are used as model aggregates to demonstrate the utility of this package of routines. The precision of the docking has allowed for the extraction of unique conformations of the macromolecules and is limited only by the reliability of the underlying structural data. Academic Press

show abstract

Section: Design Of the Situs Packagementioning

confidence: 99%

Section: Design Of the Situs Packagementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Situs: A Package for Docking Crystal Structures into Low-Resolution Maps from Electron Microscopy

Wriggers

Milligan

McCammon

1999

Journal of Structural Biology

499

492

View full text Add to dashboard Cite

show abstract

“…An efficient fast algorithm to generate codebook for vector quantization. First International Conference on Emerging Trends in Engineering and Technology, ICETET-2008, Nagpur, India, (59). Su principal limitación se encuentra en la gran complejidad computacional en tiempos de búsqueda (60,61).…”

Section: Cuantificaciónunclassified

Compresión de imágenes médicas

Noreña

Romero

2012

biomedica

View full text Add to dashboard Cite

Contribución de los autores:Ambos autores participaron en la recopilación, unificación y síntesis de los antecedentes más destacados en el área de la compresión, mediante la discusión de la información publicada sobre el tema. Además, estructuraron las metodologías ya existentes para la compresión de imágenes médicas, aportando un guía base para las personas que desean introducirse en el área. La medicina moderna es una actividad cada vez más compleja, basada en la información proveniente de múltiples fuentes: historias clínicas, dictáfonos e imágenes y vídeos provenientes de múltiples dispositivos. Las imágenes médicas constituyen una de las fuentes de mayor importancia, por cuanto ofrecen un apoyo integral del acto médico: el diagnóstico y el seguimiento. Sin embargo, la cantidad de información generada por los dispositivos de adquisición de imágenes sobrepasa rápidamente la disponibilidad de almacenamiento que tienen los servicios de radiología, lo cual genera costos adicionales en equipos de cómputo con mayor capacidad de almacenamiento. Además, la tendencia actual de desarrollo de aplicaciones en la "nube de cómputo", tiene limitaciones por cuanto, aunque el almacenamiento es virtual y está disponible desde cualquier sitio, la conexión se hace a través de internet. En estos dos casos, el uso óptimo de la información requiere necesariamente de algoritmos de compresión potentes y adaptados a las necesidades de la actividad médica. En este artículo se presenta una revisión de las técnicas de compresión más utilizadas para el almacenamiento de imágenes, así como un análisis crítico de éstas desde el punto de vista de su uso en ambientes clínicos. REVISIÓN DE TEMAPalabras clave: procesamiento de imagen asistida por computador, compresión de datos, metodologías computacionales.doi: http://dx.doi.org/10.7705/biomedica.v33i1.804 Medical image compression: a reviewModern medicine is an increasingly complex activity, based on the evidence; it consists of information from multiple sources: medical record text, sound recordings, images and videos generated by a large number of devices. Medical imaging is one of the most important sources of information since they offer comprehensive support of medical procedures for diagnosis and follow-up. However, the amount of information generated by image capturing gadgets quickly exceeds storage availability in radiology services, generating additional costs in devices with greater storage capacity. Besides, the current trend of developing applications in cloud computing has limitations, even though virtual storage is available from anywhere, connections are made through internet. In these scenarios the optimal use of information necessarily requires powerful compression algorithms adapted to medical activity needs. In this paper we present a review of compression techniques used for image storage, and a critical analysis of them from the point of view of their use in clinical settings. Las imágenes médicas son hoy en día parte fundamental de la exploración diagnóstica de un paci...

show abstract

“…Constrained search procedures include treestructured (TSVQ) (Buw, Gray, Gray, & Markel, 1980;Makhoul, Roucos, & Gish, 1985) and lattice (Gersho, 1979;Conway & Sloane, 1985) methods. In the unconstrained (full search) case, all weight vectors arc codewords, and the system searches through all the entries in a code book to find which one best represents the data.…”

Section: Vector Quantizationmentioning

confidence: 99%