Heading Homing Simulation System Based on Image Intelligent Recognition
Abstract: Channel line
recognition is one of extremely important technologies in intelligent driving
field. In recent years, machine vision has been the mainstream method to solve
channel line recognition problems. To overcome such deficiencies of existing
channel line recognition algorithms as being complicated, slow and short of
robustness, the Thesis provides a new and rapid channel line recognition
algorithm which firstly obtains outline pixel of channel line through analysis
on the images’ grayscale differences and then applies B-Spline curve to fit the
channel line profile, thus getting the final recognition effect picture. The experiments
show that excellent performances in both speed and recognition rate can result
from the algorithm. Besides, in embedded platform, the speed of the algorithm
in the Thesis results in 12 frames per second, which conforms to the real
demands of intelligent driving.
Keywords: Machine vision;
Channel line recognition; B-Spline; Curve fitting; Random sampling consistency;
Embedded system
Author: Zhao Ke, Ziba Eslami
Journal Code: jptkomputergg160115
