Classificação de regiões favoráveis a concentração de hidrocarbonetos em imagens sísmicas utilizando redes neurais artificiais

Abstract

Identifying regions favorable to the concentration of hydrocarbons (regions of interest) in seismic data is an important task, performed subjectively by geologists and geophysicists, which requires considerable time and experience. In this work an automatic process was developed to detect the regions of interest in seismic images using the Haralick descriptors and the geostatistical functions. This method is based on the use of artificial neural networks for the pattern recognition in conjunction with these descriptors. To evaluate the performance of the network, statistical metrics were used for accuracy, sensitivity, specificity, precision and F1score. The network achieved a performance of up to 98,3% accuracy and F1score, 96,7% sensitivity and 100% specificity and precision.