Summary: | 碩士 === 國立臺灣大學 === 農業機械工程學系研究所 === 86 === In this paper, a system is developed to automatically navigate a vehicle in
agriculture environment, based on the information acquired by a vision device.
The approach employs the characteristics of crops planted along linesin paral
lel, which ultimately converge at a vanishing point in a perspectiveview. Then
, we take the vanishing point as the orientation of navigation. To solve th
e discontinuity of the row crops in the image and time consuming problem encou
ntered in thinning algorithm, the study combines Houghtransform with Hotelling
transform to detect vanishing point of planted crops andcrops canopy. If the
automated vehicle is approaching the end of crop rows, the system could decide
the timing of turnaround from the acquired image. When thevehicle is moving a
long straight corridors, the vanishing point isdeterminedby extracting straigh
t line features from baseline. This paper also discussesthe effect of vari
ous mask operators on baselines detection. The determination of vanishing poin
ts and turnaround timing is sufficient tonavigate automated vehicles moving in
the environment with plantedcrops, cropscanopy or corridors from which parall
el lines could be extracted.
|