Naslov
Prepoznavanje geometrijskih svojstava prometnih znakova primjenom umjetne neuralne mreže
(Recognition of Geometrical Characteristics of Traffic Signs by the Application of Artifical Neural Network)

Autori
Gold, Hrvoje ; Kavran, Zvonko ; Kovačević, Dražen

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Peto međunarodno znanstvenostručno savjetovanje Organizacija i sigurnost prometa / Rotim, Franko - Opatija : Hrvatsko znanstveno društvo za promet, 1997, 330-334

Skup
Organizacija i sigurnost prometa

Mjesto i datum
Opatija, Hrvatska, 17.04.1997. - 18.04.1997

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
geometrijska svojstva prometnih znakova; umjetne neuralne mreže
(geometrical characteristics of traffic signs; artifical neural network)

Sažetak
One of the differences between the artifical neural networks and discrete automata, like digital computers is their ability to learn the relations between input and output symbols without need for an appropriate program. In the paper the model of the system for recognition of traffic signs based on the backpropagation learning procedure has been shown. The conditions which the system must meet in the traffic background, its structure and the way of learning and application have been also determined. The model was verified by the example of learning and recognizing the shapes of the set of traffic signs.

Izvorni jezik
Hrvatski

Znanstvena područja
Tehnologija prometa i transport

POVEZANOST RADA