THESIS
1999
viii, 87 leaves : ill. ; 30 cm
Abstract
One of the most important problems in signal processing is the extraction of one signal from other interfering signals. It has a wide range of interesting applications in our daily life. One simple method is to design a classical filter such as a bandpass filter to restore the desired signal. However, these filters are not optimal to obtain the best estimate of the desired signal. This thesis presents some other algorithms for extraction of weak signals in two biomedical applications. Some signal processing techniques such as filtering and modeling are used to enhance the signal-to-noise ratio of the measured data for extracting the desired signals....[
Read more ]
One of the most important problems in signal processing is the extraction of one signal from other interfering signals. It has a wide range of interesting applications in our daily life. One simple method is to design a classical filter such as a bandpass filter to restore the desired signal. However, these filters are not optimal to obtain the best estimate of the desired signal. This thesis presents some other algorithms for extraction of weak signals in two biomedical applications. Some signal processing techniques such as filtering and modeling are used to enhance the signal-to-noise ratio of the measured data for extracting the desired signals.
Two biomedical applications are the extraction of evoked potential in EEG signals and the determination of chemical concentration in water and urine samples. For the extraction of evoked potential, three algorithms are investigated and compared. They are commonly used ensemble averaging, adaptive filtering and our proposed linear prediction modeling coupled with ensemble averaging. Simulation results show that our proposed algorithm successfully reduce the averaging time to obtain a satisfactory waveform. For the determination of chemical concentration, three algorithms including partial least squares, hybrid linear analysis and our proposed optimal filtering coupled with partial least squares are investigated. Simulations are done on some water and urine solutions to determine the concentrations of three chemicals. The results show that our proposed method outperforms the other two algorithms in most circumstances.
Post a Comment