Unformatted text preview: d of marking the table itself,
we used a marked blanket, so we could redo this part of the process very quickly by just covering the table
with the blanket, taking the picture and removing the blanket from the table again. At the end of this part,
we had a set of samples containing inches measures correctly mapped to pixels coordinates. So we
proceeded to the next part of this task, which was to come up with a function that would, based on the
examples previously taken, predict a pair of values in inches given a pair of pixel coordinates.
We initially thought that, for this problem, the fitting curve would involve both X and Y axis,
because of the distortion created when we positioned the camera in an angular way. But after analyzing
the sample by plotting it, we noticed that both X and Y axis mappings could be achieved individually with
Linear Regression (CITATION): Using the Matlab polyfit (CITATION) function, we were able to map pixels on the image to
inches on the table. This result was used in the function that maps...
View
Full Document
 Fall '12
 ThomasTrappenberg
 Computer Science, Machine Learning, Sobel, 2 meters

Click to edit the document details