Selektivno zaznavanje in odstranjevanje plevela

diplomsko delo

Urban Kenda (Author), Suzana Uran (Mentor), Aleš Belšak (Mentor), Božidar Bratina (Co-mentor), Jurij Rakun (Co-mentor)

Abstract

V diplomskem delu smo spoznali robota Farmbeast, različne pristope k škropljenju, osnovne principe iz področja strojnega vida in uporabo robotskega operacijskega sistema. Na podlagi novopridobljenega znanja smo ustvarili sistem, ki je s strojnim vidom sposoben ločevati plevel med ozko- in širokolistnim plevelom. Rezultat prepoznave pa predstavlja vhodni podatek za novo razvito orodje, s katerim je omogočeno škropljenje z dvema različnima fitofarmacevtskima pripravkoma, glede na vrsto plevela. Za prepoznavo sta bila razvita dva različna algoritma, ki omogočata ločevanje plevela in sta bila testirana na 30 vzorcih. Test je pokazal, da prvi način v 93,3 % uspešno loči ozkolisten plevel in je 53,3 % uspešen pri ločevanju širokolistnega plevela, drugi način pa obe vrsti plevela loči 93,3 % uspešno.

Keywords

plevel;škropljenje;strojni vid;diplomske naloge;

Data

Language:	Slovenian
Year of publishing:	2020
Typology:	2.11 - Undergraduate Thesis
Organization:	UM FERI - Faculty of Electrical Engineering and Computer Science
Publisher:	[U. Kenda]
UDC:	621.865.8(043.2)
COBISS:	38393347
Views:	351
Downloads:	101
Average score:	0 (0 votes)
Metadata:

Other data

Secondary language:	English
Secondary title:	Selective detection and removal of weed
Secondary abstract:	In this work we learn about robot Farmbeast, different approaches to spraying, basic principles in the field of machine vision and the use of a robotic operating system. Based on new knowledge, we created a system which is able to separate weeds into narrow and wide-leaf sort with machine vision. The result of defining weed sort is used as input data for newly developed tool, which is capable of spraying two different phytopharmaceutical preparations, depending of the type of weed. For defining weed, two different weed separating algorithms were developed and tested on 30 samples. The test showed that first method successfully separates the narrow sort in 93,3 % and is 53,3 % successful in separating wide sort weeds. The second algorithm both of sorts separate correctly in 93,3 %.
Secondary keywords:	weeds;spraying;machine visions;
Type (COBISS):	Bachelor thesis/paper
Thesis comment:	Univ. v Mariboru, Fak. za elektrotehniko, računalništvo in informatiko, Mehatronika
Pages:	X, 52 f..
ID:	11974971