Machine Vision Davies
Machine Vision: Theory, Algorithms, Practicalities (Signal Processing and its Applications)
Davies
Usage of MV is:
- recognition problem (detect and determine characters)
- object location
2.2 Image proceessing operations
P0 is byte image variable in 3x3 window, P0 = 0..255 and P0 is in center
P4 P3 P2
P5 P0 P1 this is 3x3 window. Left shift is Q0 = P1 (new image center is P1)
P6 P7 P8
A0 bit image variable in 3x3 A0 = 0..1. Edge detection is sum neighbors
bits and if all are 1 than output is 1.
Sigma values are used for: Shrink dark objects: only if all neighbors are 1 we leave that pixel, in other words if next to a light background than expand light background (into dark object).
sigma = A1 + A2 ... + A8
if (sigma < 8) B0 = 0 else B0 = A0
Oposite, expand dark object (shring light): if next to dark object than expand dark object.
if (sigma > 0) B0 = 1 else B0 = A0
Edge finding for binary images: (if next to all dark, we mark it 0, else original)
if (sigma == 8) B0 = 0 else B0 = A0
which can we shown in table (canceling all object except near dark) sigma 0-7 8 A0 0 0 0 1 1 0 Remove salt noise (single 0 surrounded with 1) and remove pepper noise (remove binary value 1 surrounded with 0).
if (sigma<2) B0 = 0
elsif (sigma > 6) B0 = 1
else B0 = A0