Algoritmos computacionais para detecção eficiente de odontocetos em dispositivos fixos autônomos

Abstract

Considering the success of mobile devices and the evolution of its technologies, the study of Odontoceti in real time is an emerging possibility. Despite the evolution, analysis of big data chunks by complex algorithms requires considerable computing effort. Data collection of Odontoceti is execute in marine environment with limited resources, thus reducing research potential. Therefore, choosing the right algorithm to create the Workflow should maintaining a balance between computational efficiency and detection accuracy. The goal of this thesis is to propose an efficient Workflow for Passive Acoustic Monitoring. For such, a common Workflow used in the field by researchers was used as base, adding a new step for preprocessing of captured data. The detection step, focus of this thesis due to its applicability and notable impact on the next steps, is responsible to analyse received acoustic signals, filtering a good amount of data. The next step condensates data in a way that facilitates transfer of captured information to remote locations. In sequence there is the step responsible for the identification of received information from the previous step. The last step is based on software components to study relevant information. The appplication of the detection step have shown a satisfactory performance providing a reduction of 96.52 of total data to be processed, making it easy for relevant information to be identified and distributed online to remote research stations.