Publication:
Reducción de ruido en señales de espectroscopia raman
Reducción de ruido en señales de espectroscopia raman
Authors
Quintero Pizo, Luis A.
Embargoed Until
Advisor
Hunt, Shawn D.
College
College of Engineering
Department
Department of Electrical and Computer Engineering
Degree Level
M.S.
Publisher
Date
2008
Abstract
En este documento se desarrollan algoritmos de reducción de ruido para mejorar señales de espectroscopia Raman manteniendo la información de la señal. Dos tipos de ruido afectan las medidas, ruido impulsivo causado por radiación de alta energía y rayos cósmicos, y ruido aleatorio producido en la estimación de la dispersión Raman. Dos etapas secuenciales fueron desarrolladas para procesar imágenes Raman; la primera etapa trata de remover ruido impulsivo y la segunda reduce el ruido aleatorio. Los algoritmos fueron probados con datos reales, así como datos sintéticos, y comparados usando el promedio del cuadrado de los errores (MSE por sus siglas en ingles) y la norma infinita (L∞). Con relación a la reducción de ruido aleatorio, los resultados muestran un mejor comportamiento para eliminación de ruido con Wavelets, en términos de MSE y L∞, en comparación con los filtros Kalman y el ampliamente usado filtro de Savitzky-Golay.
Noise reduction algorithms were developed in order to improve Raman spec- troscopy signals while preserving signal information. Two kinds of noise affect the measurements, impulsive noise caused by high-energy radiation and cosmic rays, and random noise produced in Raman backscattering estimation. Two sequential stages were developed to process the Raman images; the first stage attempts to remove the impulsive noise and the second one reduces the random noise. The explored algorithms have been tested using real data as well as synthetic data, and compared using the Means Squared Error (MSE) and the Infinity Norm (L∞). In terms of random noise reduction, results show Wavelet denoising performs better, in the sense of MSE and L∞, compared to Kalman filters and the widely used Savitzky-Golay filter.
Noise reduction algorithms were developed in order to improve Raman spec- troscopy signals while preserving signal information. Two kinds of noise affect the measurements, impulsive noise caused by high-energy radiation and cosmic rays, and random noise produced in Raman backscattering estimation. Two sequential stages were developed to process the Raman images; the first stage attempts to remove the impulsive noise and the second one reduces the random noise. The explored algorithms have been tested using real data as well as synthetic data, and compared using the Means Squared Error (MSE) and the Infinity Norm (L∞). In terms of random noise reduction, results show Wavelet denoising performs better, in the sense of MSE and L∞, compared to Kalman filters and the widely used Savitzky-Golay filter.
Keywords
Usage Rights
Persistent URL
Cite
Quintero Pizo, L. A. (2008). Reducción de ruido en señales de espectroscopia raman [Thesis]. Retrieved from https://hdl.handle.net/20.500.11801/2355