Developpement de techniques utilisant les ondelettes pour l’analyse,

Abstract

n this thesis, we tried to address several topics concerning the analysis, synthesis and compression of the electrocardiogram (ECG) signal using the MIT database. Where we have detected R waves by locating each QRS complex using the dyadic wavelet transform, which defines a window of R peak search, and then, we have accurately detected the R waves by the manipulation of the first order coefficients of the discrete wavelet transform. In the second part of our work we have segmented the signal in its RR periods in order to construct vectors of a data matrix, where we extract its principal components in order to reduce the size of the information, then, we reduce more and more the size by applying a threshold on the signal resulting from the discrete cosine transform. The last algorithm designed is a practical application of the R wave detection algorithm, but, after the extraction of one-dimensional signal of the resulting scanned image of the ECG paper. The results of R peaks detection, reducing the size of the signal and the correspondence of samples of the signal resulting from the extraction with those of the original file, was very satisfactory and with an extreme precision, demonstrating the quality of our algorithms presented in this thesis.