Unformatted text preview: ext phase,
we converted the pixel value of the red ball to square coordinates on the table.
For the green block detection what differs from the red ball detection is that we used a bigger filter. We
used a 5 by 5 matrix of ones as our green block detector filter. Instead of using the red dimension of the
RGB 3 dimensional matrix of the raw image, we used the green dimension, of course. Then, after
convolute the matrix with our filter, we used, again the Matlab find function, but this time to find all the
pixels that had value higher than the mean of the values of the convoluted matrix. We added 10 to the
mean before compare to the pixels values in order to raise the threshold making that only greener things of
the image are considered. Analogously with the red ball detection approach, we finally get all the pixel's
coordinates found and convert them into square coordinates.
For the path finding: A* as it was provided on the professor’s website removing the borders to avoid the
robot from falling over the table.
In order to ensure better navigation and path planning for the robot, we employed A* algor...
View Full Document
This document was uploaded on 02/14/2014 for the course CSCI 6505 at Dalhousie.
- Fall '12
- Machine Learning