Pohitritev transformacije domnevnih razdalj

magistrsko delo

Danijel Žlaus (Author), Domen Mongus (Mentor)

Abstract

V magistrskem delu opisujemo pohitritev transformacije domnevnih razdalj, ki je izpeljanka tradicionalnih algoritmov transformacij razdalj. Transformacije razdalj običajno delujejo nad dvodimenzionalnimi binarnimi slikami, kjer vsakemu elementu ospredja določijo oddaljenost do najbližjega elementa ozadja. Kadar slika ni binarna, je nad njo potrebno izvesti dodano predprocesiranje, ki vključuje korak binarizacije. Nasprotno pa lahko transformacijo domnevnih razdalj uporabimo neposredno nad sivinskimi, barvnimi in multispektralnimi slikami in se tako izognemo pogoste neželenemu predprocesiranju. Slabost tega pristopa pa je časovna zahtevnost, ki je v naivni implementaciji kar O(N^2.5). V magistrskem delu predstavimo pohitren algoritem transformacije domnevnih razdalj ter teoretično analizo njegove časovne zahtevnosti. Nad implementiranim algoritmom izvedemo tudi meritve, s čimer potrdimo teoretične časovne zahtevnosti pohitrenega pristopa, ki je enaka O(N^1.5) v pričakovanem ter O(N^2) v najslabšem primeru.

Keywords

matematična morfologija;transformacija domnevnih razdalj;transformacija razdalj;časovna zahtevnost;optimizacijske metode;magistrske naloge;

Data

Language:	Slovenian
Year of publishing:	2016
Typology:	2.09 - Master's Thesis
Organization:	UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher:	D. Žlaus
UDC:	004.93'11(043.2)
COBISS:	19836182
Views:	1007
Downloads:	124
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Optimization of Quasi Distance Transform
Secondary abstract:	This thesis presents an optimisation of Quasi Distance Transform, derived from traditional distance transformation algorithms. Distance transform is typically applied on two-dimensional binary images, where each foreground element is assigned the distance to its closest background element. When the image is not binary, additional preprocessing with binarisation is required. On the other hand, Quasi Distance Transform can be applied directly to grayscale, colour or multi-spectral images, thus avoiding often unwanted preprocessing. However, the weakness of the method is its time complexity, which is in naive implementation equal to O(N^2.5). In this thesis, we present a new optimised algorithm and the theoretical analysis of its time complexity. We confirm the theoretical time complexity for quasi distance transformation together with measurements, thus proving they are N^(1.5) for the expected and O(N^2) for the worst case.
Secondary keywords:	quasi distance transform;distance transform;mathematical morphology;time complexity;optimisation methods;
URN:	URN:SI:UM:
Type (COBISS):	Master's thesis/paper
Thesis comment:	Univerza v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Računalništvo in informacijske tehnologije
Pages:	XII, 35 f.
ID:	9161589