Processamento de imagens distribuído usando o Hadoop e HIPI

Abstract

Distributed data processing is an important technology that grew up in the last years due mainly to the increase in quantity of information produced by the Internet. For this type of processing get satisfactory levels, it is necessary to use appropriate computer tools (hardware and software). Hadoop is a great example of software used for this end, and it has being used in computer clusters to create an environment that allows distributed data processing. For images processing, Hadoop uses a specialized extension called HIPI (Hadoop Image Processing Interface), which provides execution models based on known processing paradigms and imaging. In this work, tests were performed using Hadoop and HIPI, to determine the advantages and disadvantages of using distributed processing in a specific technique of digital image processing. For testing, it was applied a known and used technique for background modelling. Also, it was adapted to operate in a distributed environment. The results obtained demonstrate the superiority of the distributed processing using Hadoop and HIPI in relation to the sequential processing.