07.11.2014 Views

Application and Optimisation of the Spatial Phase Shifting ...

Application and Optimisation of the Spatial Phase Shifting ...

Application and Optimisation of the Spatial Phase Shifting ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

86 Electronic or Digital Speckle Pattern Interferometry<br />

3.4.4 <strong>Spatial</strong> phase shifting on speckle fields<br />

We have seen in Fig. 3.21 <strong>and</strong> Fig. 3.22 that <strong>the</strong> positive <strong>and</strong> negative interference b<strong>and</strong>s overlap exactly<br />

in <strong>the</strong> spatial frequency plane when <strong>the</strong> source point <strong>of</strong> <strong>the</strong> reference wave coincides with <strong>the</strong> centre <strong>of</strong> <strong>the</strong><br />

aperture. As <strong>the</strong> origin <strong>of</strong> <strong>the</strong> reference wave is laterally displaced, <strong>the</strong> overlap <strong>of</strong> <strong>the</strong> interference spectra<br />

gets smaller; ideally <strong>the</strong>y can be fully separated in <strong>the</strong> frequency plane, as shown in Fig. 3.29 for two<br />

different settings <strong>of</strong> phase shift <strong>and</strong> speckle size. In <strong>the</strong> Fourier formalism, <strong>the</strong> carrier frequency manifests<br />

itself as a constant phase factor, which shifts <strong>the</strong> interference spectra by ν c , with ν c being <strong>the</strong> spatial<br />

carrier frequency, <strong>and</strong> thus turns <strong>the</strong>m into <strong>the</strong> so-called signal sideb<strong>and</strong>s. We defer a more detailed<br />

discussion to Chapter 6.5.<br />

−ν N<br />

0<br />

−ν N<br />

0<br />

ν y<br />

ν y<br />

ν N ν N<br />

–ν N 0 ν x ν N –ν N 0 ν x ν N<br />

Fig. 3.29: Power spectra (log scale) <strong>of</strong> speckle interferograms with carrier frequency; left: α x =120°/column<br />

(ν c,x =1/(3 d p )), d s =3.5 d p ; right: α x =90°/column (ν c,x =1/(4 d p )), d s =2.5 d p . To allow for sufficient ∆x to<br />

obtain α x =120°/column, <strong>the</strong> fibre end is in a slit beside <strong>the</strong> aperture (cf. Fig. 5.1); to <strong>the</strong> right, ∆x D/2,<br />

<strong>and</strong> <strong>the</strong> fibre guide obscures part <strong>of</strong> <strong>the</strong> aperture. The contrast <strong>of</strong> <strong>the</strong> images has been enhanced to make<br />

<strong>the</strong> speckle halo visible.<br />

The width <strong>of</strong> <strong>the</strong> side b<strong>and</strong>s in an interferogram's power spectrum indicates <strong>the</strong> range <strong>of</strong> speckle phase<br />

gradients that distort <strong>the</strong> carrier fringes. As already hinted in 2.2.3.2, <strong>the</strong>se distortions are equivalent to<br />

local miscalibrations <strong>of</strong> <strong>the</strong> phase shift, which makes great dem<strong>and</strong>s on <strong>the</strong> miscalibration tolerance <strong>of</strong> <strong>the</strong><br />

phase-reconstruction formula. Also, its spectral response should utilise as much <strong>of</strong> <strong>the</strong> signal as possible;<br />

but as we have seen in 3.2.2, nei<strong>the</strong>r is easy to be had.<br />

Complete separation <strong>of</strong> <strong>the</strong> interference b<strong>and</strong>s is desirable because <strong>the</strong>n all frequency components <strong>of</strong> <strong>the</strong><br />

signal will be unambiguous. If α x is to have <strong>the</strong> same sign throughout <strong>the</strong> interferogram, one has to<br />

dem<strong>and</strong> that <strong>the</strong> positive/negative signal frequencies occupy no more than <strong>the</strong> positive/negative halfplane,<br />

(ν x+ ,ν y ) <strong>and</strong> (ν x– ,ν y ), in <strong>the</strong> frequency spectrum. If <strong>the</strong>se boundaries are crossed, <strong>the</strong> signal b<strong>and</strong>s<br />

will overlap around ν x = 0, or with aliasing (see below) around ν x =ν N , or both. We will consider<br />

examples <strong>of</strong> such power spectra in Chapter 5.5.3.<br />

However, it is possible to permit sideb<strong>and</strong>s larger than in Fig. 3.29 on <strong>the</strong> right <strong>and</strong> still avoid <strong>the</strong>ir<br />

mixing when we record information in <strong>the</strong> ν y co-ordinates as well <strong>and</strong> thus truly utilise <strong>the</strong> 2-D nature <strong>of</strong><br />

<strong>the</strong> measurement. Depending on <strong>the</strong> speckle size <strong>and</strong> shape, <strong>the</strong>re may <strong>the</strong>n be various solutions to

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!