Russell Kirmayer, HW 4. Images after convolution
The masks are good for edge detection in the sense that they are essentially taking the derivative across the pixels from left to right, or in other words are taking the average of the pixels.
A smaller mask is more precise and is faster to compute, and a larger mask is better for larger edges that a smaller mask would skip over.
-1 -1 -1; 0 0 0; 1 1 1. This mask would be good for up-down averaging, like the horizon
Original Image 1
Image 1, mask 1
Image 1, mask 2
Image 1, mask 3
Original Image 2
Image 2, mask 1
Image 2, mask 2
Image 2, mask 3
Convolution Lab directory Source Code Line Following Demonstration
Line Following DemonstrationDead Reckoning