Obrada sonarske slike u svrhu detekcije šake ronioca

master's thesis

• Title (english): Sonar image processing for the purpose of diver hand detection
• Author: Franka Guštin
• Mentor: Nikola Mišković (mentor)
• Committee member: Nikola Mišković (predsjednik povjerenstva)
• Committee member: Zoran Vukić (član povjerenstva)
• Committee member: Marija Seder (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2015-07-17, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering

Abstract

Ronioci prilikom svojih misija djeluju u opasnom i slabo nadziranom okruženju gdje i najmanji neočekivani tehnički kvarovi ili nedostatak pažnje mogu dovesti do katastrofalnih posljedica. Iz tog se razloga koriste autonomne podvodne ronilice koje pomažu roniocu na misijama i prate njegovo fizičko i psihičko stanje. U radu je implementiran sustav koji na temelju sonarske slike detektira šaku ronioca i određuje broj ispruženih prstiju te tako omogućava interakciju između ronioca i ronilice. S obzirom na ograničenja uzrokovana kvalitetom sonarske snimke, za detekciju šake ronioca odabran je algoritam Viole i Jonesa koji se temelji na Haarovim značajkama. Drugi dio zadatka predstavlja određivanje broja ispruženih prstiju, te su za tu svrhu implementirana dva algoritma: stroj s potpornim vektorima i metoda konveksne ljuske. Algoritmi su testirani na dostupnim sonarskim snimkama.

Abstract (english)

Divers operate in harsh and poorly monitored environments in which the slightest unexpected technical malfunction or lack of attention can result in catastrophic consequences. To overcome this problem, autonomous underwater vehicles are used, which help divers on underwater missions and monitor their physical state and actions. To achieve interaction between diver and autonomous vehicle, hand detection and gesture classification system based on sonar image is implemented. Due to the limitations caused by the quality of sonar image, for hand detection is chosen Viola and Jones algorithm based on Haar features. For gesture classification problem, two algorithms are implemented: support vector machine and convex hull method. Algorithms are tested on available sonar recordings.

• Keywords: sonarska slika; detekcija šake; prepoznavanje gesti; kaskada Haarovih klasifikatora; Hu momenti; stroj s potpornim vektorima; konveksna ljuska; konveksni defekti
• Keywords (english): sonar image; hand detection; gesture classification; cascade of boosted Haar classifiers; support vector machine; Hu moments; convex hull; convexity defects
• Language: croatian
• URN:NBN: urn:nbn:hr:168:045188
• Study programme: Title: Electrical Engineering and Information Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije (magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-06 09:06:29