This preview shows page 1. Sign up to view the full content.
Unformatted text preview: e look
smaller than the nearest one, we had to adopt a strategy to correct it.
Initially we measured the table and decided how many inches a side of a square would have. We
made it in a way that the table would be a grid of 8 by 8 squares, as the table was a square itself. This measured squares of 9x9 inches each. We delimited such squares virtually, with a routine that would give
us the index of the square (x and y positions), given a pair of measures in inches.
After all the physical measurements, we started the mapping process. Our objective in this part
was to pass the pixel coordinates to a function and obtain coordinates of the corresponding square on the
table, but as we already had a function that mapped inches to square, mentioned in the above paragraph,
all we had to do was to map pixels to inches.
In order to map pixels to inches, we decided to mark different parts of the table, take their
measures in inches, take a picture of the table and manually assign each marking to the corresponding
pixel on the image. We came across the problem of the shared environment in this part: As many groups
were using the same table as we were, in order to minimize disturbance, instea...
View Full
Document
 Fall '12
 ThomasTrappenberg
 Machine Learning

Click to edit the document details