Fondazione Bruno Kessler - Technologies of Vision

contains material from
Template Matching Techniques in Computer Vision: Theory and Practice
Roberto Brunelli © 2009 John Wiley & Sons, Ltd

2.2 Diffraction Effects

When light passes through a small aperture, i.e. an aperture whose size if comparable to the wavelength of light, the wave nature of light causes interference effects that result in perceivable image quality loss. Images become more blurred due to the increasing support of the resulting point spread function of the optical system. An immediate consequence is that we cannot indefinitely gain depth of focus by reducing the aperture of the optical system as Equation TM:2.22 would suggest.

The following code snippet simulates the effect of diffraction with data corresponding to a high resolution digital compact camera, assuming a default wavelength of 550e- 9m: diffraction is modeled by convolution with an appropriate kernel

1 A  <- ia.get(i_orthoRed, animask(0,0,1201,256)) 
2 # 
3 # dK represents the points spread function of the 
4 # optical system (and corresponds to the Aity function) 
5 # 
6 dK <- tm.diffractionKernel(focalLength= 8.9e-3, 
7 ...                            fStop=       16, 
8 ...                            pixelSize=   (8.8/3264)/1000) 
9 tm.plot("figures/diffractionKernelF16", 
10 ...  persp(dK, main = "Airy diffraction kernel at f16")) 
11 # 
12 # In order to better appreciate the shape of the kernel we 
13 # can oversample it: dK3 can then not be used to simulate 
14 # the effect (as it corresponds to 3x oversmoothing) 
15 # 
16 dK3 <- tm.diffractionKernel(focalLength= 8.9e-3, 
17 ...                             fStop=       16, 
18 ...                             pixelSize=   (8.8/3264)/1000, 
19 ...                             superSampling=3) 
20 tm.plot("figures/diffractionKernelF163x", 
21 ...  persp(dK3, main = "Airy diffraction kernel at f16 (3x supersampled)")) 
22 # 
23 # We generate the diffracted image 
24 # 
25 B <- ia.convolution(A, dK) 
26 tm.plot("figures/diffractionEffect", 
27 ...           ia.show(B, main="Diffraction at f16")) 
28 # 
29 # and have a detailed look at an intermediate row 
30 # 
31 tm.plot("figures/diffractionDetail", 
32 ...          matplot(cbind(A@data[126,],B@data[126,]), 
33 ...                  type="l",col=c("black"), pch = 2:3, lty=1:2, xlab="x", 
34 ...                  ylab="", main="W(/o) Diffraction"), 
35 ...          legend(0,150, c("No diffraction", "Diffraction"), 
36 ...                 lty=1:2, col=c("black")), 
37 ...          height=3)


PIC PIC
PIC

Figure 2.3: Diffraction results in significant contrast loss at commonly encountered operating conditions.