Analiza učinkovitosti rekonstrukcije različitih transformacija kod sažimajućeg očitavanja u svrhu reprezentacije slike

master's thesis

• Title (english): Analysis of Reconstruction Efficiency of Different Transforms for Image Representation Using Compressive Sensing
• Author: Maja Barić
• Mentor: Davor Petrinović (mentor)
• Committee member: Davor Petrinović (predsjednik povjerenstva)
• Committee member: Damir Seršić (član povjerenstva)
• Committee member: Mladen Vučić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2017-07-07, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering

Abstract

U sklopu ovog diplomskog rada proučen je način rada sažimajućeg očitavanja te transformacije koje imaju potencijal korištenja u sklopu tog algoritma pri obradi slike. Proučen je utjecaj svake od transformacija na kvalitetu rekonstrukcije slike. Došlo se do zaključka da Huang – Hilbertova transformacija nije pogodna za upotrebu u sklopu ovog algoritma. Nadalje, KLT transformacija daje puno bolje rezultate kao mjerna matrica nego matrica transformacije. Konačno, DCT i DWT se koriste u ulozi matrice transformacije. Zaključeno je da je bolje raditi transformaciju nad blokovima slike nego nad cijelom slikom zbog skraćivanja izvođenja rekonstrukcije slike.
I DCT i DWT imaju svoje prednosti i mane. Kod DCT može se postići prikaz slike za mnogo manji postotak zadržanih uzoraka u njenom rijekom prikazu. S druge strane, kod rekonstrukcije slike DWT ipak daje nešto bolje rezultate. Obrada slike je, bez sumnje, područje u kojem sažimajuće očitavanje ima veliki potencijal poboljšanja trenutno korištenih sustava.

Abstract (english)

This thesis includes a study of compressive sensing and various image transforms which have a potential of being used in that algorithm in image processing. Each of four studied transforms has a unique effect on quality of reconstructed image.
A few conclusions were made. Huang - Hilbert transform cannot be used as a transformation matrix in compressive sensing. Furthermore, KLT is more suitable in the role of measurement matrix than transformation matrix. Finally, DCT and DWT are used as transformation matrices. It was also deducted that processing image blocks is more effective than processing image as a whole.
Both DCT and DWT have some good and bad properties. DCT enables higher image sparsity rate (it is sufficient to preserve 20% of coefficients) than DWT. On the other hand, DWT offers better perceptive quality of reconstructed image. Image processing is, without any doubt, area in which compressive sensing has a high potential of improving currently used systems.

• Keywords: sažimajuće očitavanje; diskretna kosinusna transformacija; diskretna valićna transformacija; Karhunen – Loève transformacija; Huang – Hilbertova transformacija; kvaliteta rekonstruirane slike
• Keywords (english): compressive sensing; discrete cosine transform; discrete wavelet transform; Karhunen - Loève transform; Huang - Hilbert transform; quality of reconstructed image
• Language: croatian
• URN:NBN: urn:nbn:hr:168:767129
• 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: 2019-03-13 18:51:28