Recuperación de señales dispersas utilizando orthogonal matching pursuit (omp)
Loading...
Authors
Lobato Polo, Adriana Patricia
Ruiz Coral, Rafael Humberto
Quiróga Sepúlveda, Julián Armando
Recio Vélez, Adolfo León
Director
Content type
Artículo de revista
Document language
EspañolPublication date
2009
Journal Title
Journal ISSN
Volume Title
PDF documents
Abstract
Muestreo compresivo es una rama emergente del procesamiento de señales, basada en el hecho de que un número pequeño de proyecciones lineales no adaptativas sobre una señal compresible contiene suficiente información para reconstruirla y proce- sarla. En este artículo se presentan los resultados obtenidos al evaluar cinco matrices de medición para la realización de mues- treo compresivo en un sistema que utiliza el algoritmo orthogonal matching pursuit (OMP), para la recuperación de la señal ori- ginal. Las matrices de medición están implicadas tanto en el proceso de muestreo–compresión de la señal, como en la recons- trucción de la misma. Dentro de este grupo de matrices estudiadas se destacó la matriz Hadamard aleatoria, con la cual es po- sible obtener el menor porcentaje de error en la recuperación de la señal. Adicionalmente se presenta una metodología para la evaluación de estas matrices, que permita posteriores análisis de la idoneidad de estas para aplicaciones específicas.
Compressive sensing is an emergent field of signal processing which states that a small number of non-adaptive linear project- tions on a compressible signal contain enough information to reconstruct and process it. This paper presents the results of e- valuating five measurement matrices for applying them to compressive sensing in a system using orthogonal matching pursuit (OMP) to reconstruct the original signal. The measurement matrices were those implicated in compressive sensing as well as in reconstructing the signal. The Hadamard-random matrix stood out within this group of matrices because the lowest percentage of error in signal recovery was obtained with it. This paper also presents a methodology for evaluating these matrices, allowing sub- sequent analysis of their suitability for specific applications.
Compressive sensing is an emergent field of signal processing which states that a small number of non-adaptive linear project- tions on a compressible signal contain enough information to reconstruct and process it. This paper presents the results of e- valuating five measurement matrices for applying them to compressive sensing in a system using orthogonal matching pursuit (OMP) to reconstruct the original signal. The measurement matrices were those implicated in compressive sensing as well as in reconstructing the signal. The Hadamard-random matrix stood out within this group of matrices because the lowest percentage of error in signal recovery was obtained with it. This paper also presents a methodology for evaluating these matrices, allowing sub- sequent analysis of their suitability for specific applications.