Očitavanje štampanog teksta pomoću metoda strojnog učenja

master's thesis

• Title (english): Reading printed text using machine learning methods
• Author: Bruno Bandić
• Mentor: Marko Čupić (mentor)
• Committee member: Marko Čupić (predsjednik povjerenstva)
• Committee member: Jan Šnajder (član povjerenstva)
• Committee member: Siniša Šegvić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-14, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Opisujemo proces očitavanja štampanog teksta pomoću metoda strojnog učenja. Prvo prolazimo kroz binarizaciju dokumenta. Onda prolazimo kroz segmentaciju tog binariziranog dokumenta. Onda prolazimo kroz procesiranje tog binariziranog dokumenta gdje procesiramo sliku dokumenta tako da mičemo šum i pripremamo ga za izdvajanje značajki. Opisujemo razne metode izdvajanja značajki, te kako sa svakom drugačije opisujemo znakove. Opisujemo razne metode raspoznavanja, te kako svaka od metoda raspoznavanja uzima značajke, te iz njih pokušava raspoznati o kojim se to znakovima radi. Te još pobliže opisujemo metode strojnog učenja. Prolazimo kroz rezultate raspoznavanja te opisujemo kako postaviti parametre neuronskih mreža da bi bolje funkcionirale.

Abstract (english)

We describe the proces of reading printed text using machine learning methods in general. Firstly we go through the binarization of the document. We then go through the segmentation of that document by removing the noise and preparing it for feature extraction. We describe various methods of feature extraction, and how we use those methods to represent symbols. We describe various methods of recogniton, and how each of those methods takes the features of each symbol and try and recognize what symbols there are. We closely describe methods of machine learning. We go through the results of recogniton and describe how to set the neural network paramaters so they could work better.

• Keywords: očitavanje štampanog teksta; binarizacija; segmentacija; predprocesiranje; morfološke operacije; izdvajanje značajki; raspoznavanje; strojno učenje; neuronske mreže; postprocesiranje
• Keywords (english): reading printed text; binarization; segmentation; preprocessing; morphological operations; feature extraction; pattern recognition; machine learning; neural networks; postprocessing
• Language: croatian
• URN:NBN: urn:nbn:hr:168:084133
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-14 08:23:08