Algoritmo de compresión de EEG mediante predicciones usando el polinomio de interpolación de newton con diferencias divididas

Abstract

This paper presents an on-line algorithm for compressing EEG signals with low computational cost, ideal for microcontroller applications. A Newton Divided Differences Interpolation Polynomial is used. An amount of 98 experiments were made for testing the method. Parameters as Compression Rate, Correlation Coefficient and Mean Square Distortion were calculated in order to assess the compression quality. The lowest Compression Rate measured was 37.60% while the lowest Correlation Coefficient was 99.52% and the highest Mean Square Distortion was 6.872%. © Springer-Verlag Berlin Heidelberg 2007.