OPTICAL CHARACTER RECOGNITION ON GRID AND MULTI-CORE SYSTEMS - PERFORMANCE ANALYSIS

Requirements of digital libraries for computing power, as well as needs of smaller users for automated document processing and digitization are growing, making grid environments and multi-core systems a preferred platform for optical character recognition applications. This paper examines character recognition performance on the CRO-NGI grid and on two multi-core systems by using from a single core up to all available cores. In addition, different models of load distribution on processing units are used. Comparison of the processing speed and the recognition accuracy is conducted. A developed optical character recognition application for parallel environments is used to obtain the results.Original Abstract: Zahtjevi digitalnih biblioteka za racunalnom moci, kao i potrebe manjih korisnika za automatiziranom obradom dokumenata i digitalizacijom su u porastu, sto cini grid i visejezgrena okruzenja pozeljnom platformom za aplikacije optickog prepoznavanja znakova. U clanku se ispituju performanse prepoznavanja znakova na CRO-NGI gridu i na dva visejezgrena racunala, uz koristenje od jedne do svih dostupnih jezgara. Uz to, koristeni su razliciti modeli raspodjele opterecenja na procesne jedinice. Provedena je usporedba brzine obrade i tocnosti prepoznavanja. Za dobivanje rezultata koristena je razvijena aplikacija optickog prepoznavanja znakova za paralelno okruzenje.