Automatic Seismic Signal Detection via Record Segmentation

摘要: The automatic seismic signal detection constitutes a very interesting and challenging task. main difficulty in solving this problem is attributed to the fact that both statistical properties of noise characteristics recorded events are general unknown. In paper, by exploiting particular nature signals we treating, as well number possessed exchangeable random variables, formulate at hand record segmentation one propose use two functionally linked test statistics for its efficient robust solution, two-step procedure. By following approach, succeed not only detecting wave arrival, but also identifying entire interval occupied signal, while minimizing required parameters. performance proposed technique confirmed series experiments, synthetic real data sets.