ESPI: A Tutorial


Giuseppe SchirripaSpagnolo received a degree cum laude in Physics, in 1978from the Università della Calabria. From 1979 to 1984 he wasHead of the R&D Section of Strumentazione Elettronica Avanzatain Rome. In 1985 he joined the Engineering Faculty of theUniversity of L'Aquila as an Assistant Professor. In 1986 he wasvisiting researcher at the University of Wisconsin—Madison. In1998 he joined the Electronic Engineering Department of theUniversity of Roma Tre as an Associate Professor. His researchinterests include electronic instrumentation and sensors, opticalmetrology, and digital image processing. In these fields he haspublished more than 90 papers. He is a member of SPIE, OSA, andIEEE-LEOS.

Electronic Speckle Pattern Interferometry(ESPI) is a very effective optoelectronic system fornon-destructive testing. This technique allows us to measure thefull-field surface deformation of a scattering object. ESPI hasbeen used in many fields, including the study and characterizationof building materials, artwork conservation assessment, andapplications in the automotive industry.

An interesting feature of the ESPI system is the possibility ofgrabbing frames continuously from a deforming object andsubtracting them in succession from a reference image. You can usethis technique to observe the real-time evolution of the fringepattern, related to the deformation of the surface you areinvestigating. ESPI produces results in the form of images (fringepatterns). You need to use digital image-processing techniques tomanipulate and process the results to perform a quantitativeevaluation.

The Development of Non-DestructiveTesting

Material characterization and structureevaluation is a fundamental task in many fields. Experimentalmeasurement methods should have a high sensitivity withoutaffecting, with their presence, the results. This is particularlytrue when small-size specimens are studied.

Optical methods are inherently non-intrusive and non-contacting.Furthermore, they are highly sensitive and provide full-fieldmeasurement data. After the advent of the laser in 1960, it tookabout a decade until the introducion of visual-data electronicprocessing in interferometry.

Laser and holographic interferometry are now well-establishedtechniques in optical non-destructive testing. However,shortcomings such as stringent stability requirements, high costs,and the need for optically skilled operators hindered the spread ofthese methods in hostile and/or industrial situations. In the1980s, ESPI received a decisive boost with the arrival ofcomputerized fringe analysis and the dramatic development of CCDs,electronic cameras, frame grabbers, and personal computers.

Figure 1:  Non-destructive vs. destructivetesting

ESPI Basics

Speckles are the bright and dark spotsthat appear on a diffuse scattering surface when illuminated bylaser light. This phenomenon occurs because waves of laser lightscattered by various surface elements are superimposed with arandom phase relationship in space.

Speckles were initially considered as an additional, unwanted,side effect of laser light, and considerable effort was devoted tofind methods for reducing this noise. Then scientific communitythen turned its attention to the harnessing of the speckle effect.This led to the development of a new branch of optical techniques,called speckle metrology, now firmly established in non-destructivetesting and inspection.

Consider ESPI as a form of image holography (which is simply ahologram recorded of the object instead of the object itself) withan in-line reference beam. The CCD camera replaces the plate as arecording medium and electronic processing performs thereconstruction.

There are two types of fundamentally different specklepatterns:

  • Photographic speckle patterns contain information onlyabout light-wave amplitude. The patterns are obtained when adiffusely scattering surface of an object is illuminated bycoherent light and the resulting scattered wavefront is recorded ona photographic emulsion.
  • Holographic speckle patterns contain information aboutboth amplitude and wave phase. The patterns are obtained when thescattered wavefront emanating from the diffuse surface is recordedsimultaneously with a reference wavefront on the photographicemulsion.

The basic principle of ESPI is the recording of a holographicspeckle-pattern sequence on the photosensor of a TV camera. This isthe main advantage that distinguishes ESPI from holography. Therecording step is identical to that of image holography. However,since the CCD has lower resolution than holographic films, the CCDmust resolve the holographic speckle pattern details (such asspeckles).

The CCD camera's image is converted into a corresponding videosignal. This video signal is electronically processed through anintermediate recording medium (commonly a frame grabber), so thattexture variations of the speckle pattern are converted intobrightness variations. A speckle interferogram is generatedarithmetically by subtracting two digitized speckle patterns.

In practice, the intensity distribution in the detector plane isstored with the object in its reference state. The object is thendeformed and a second frame is stored. The two frames are thensubtracted and correlation live fringes are displayed on a monitor(Figure 2 ).

Figure 2:  Schematic of ESPI measurement showing howyou can use a monitor to display subtracted and correlationfringes

The resulting fringes are similar in appearance to conventionalholographic fringes, but with a lower image quality, due to muchmore evident speckle noise. For this reason, ESPI fringes areusually digitally treated for noise removal and contrastenhancement.

Since you can consider each processed video frame to be arecorded and reconstructed hologram, 25 holograms (by the EuropeanTV standard) are presented on the TV monitor each second. Thanks tothis high-repetition rate, combined with the relatively shortexposure time of 40 ms (European TV standard), the technique is notaffected by the instabilities that would destroy most conventionalinterferometric holograms.

If a squared difference is performed between two digitizedspeckle patterns and recorded at different states of the object,the result will be:

Io (x,y ) andIr (x,y ) are the object and reference beamintensities, and (x,y ) is the speckles random phase. The phase variationbetween the two subtracted patterns is contained in (x,y ). Equation 1 represents a typical ESPIinterferogram obtained by subtraction technique. It is afflictedwith speckle noise, appearing as bright and dark spots,superimposed on the fringe pattern (the true image). Usually, thefringe spacing is large compared with the mean speckle size. Youcan thus treat speckle noise as high-frequency noise, which you canreduce with low-pass filtering.

In particular, taking an average operation over many specklesand considering the statistical properties of the speckle pattern,we obtain:

Equation 2 is equivalent to the classical interferometryequation. It describes perfect correlation fringes withoutspeckles. In practice, due to speckle decorrelation, fringevisibility is always <>

Fringe visualization is sufficient for qualitative evaluation innon-destructive testing. For quantitative evaluation, you must usedigital-fringe analysis techniques to numerically extract (phaseretrieval) the phase information (x,y ).

ESPI Architectures

You can describe basic ESPI architecturesusing functional blocks, for example, with Digital Vision's Witprogramming software. Digital Vision designed Wit for easyintegration of C functions, data structures, and parallelexecution. A functional block in Wit is an operator, which receivesdata objects, carries out some processing, and then produces newdata objects.

Figure 3:  ESPI descriptions in terms of Witfunctional blocks

Figure 3 is not only an ESPI block diagram but is also analgorithm, since you can load and run it from Wit. After starting,a (first) image of the object in its reference state is acquiredand stored. Then images (imm) recorded at different states of theobject are acquired and digitally subtracted (diff) from thereference image. The resulting interferogram is displayed on amonitor in quasi-real time. You can also store (save) theinteferogram on disk and use it for quantitative analysis.

In fact, we can retrieve the desired information performing aFourier-based demodulation. First the image is fast Fouriertransformed (FFT). Then, filtering in the frequency plane, weselect a side lobe and shift it into the origin (filtering). Aninverse FFT (IFFT) yields a complex image from which you can obtainthe phase (arctan). This phase is wrapped in the range , therefore the final step involves a phase unwrappingprocedure.

Figure 4:  A typical ESPI arrangement for anin-plane sensitive interferometer. Coherent light from a laser iscoupled to a single-mode fiber, amplitude divided, and spatiallyfiltered into two beams of equal intensity, which illuminate theobject at equal and opposite angles.

Figure 5:  A typical ESPI arrangement for anout-of-plane sensitive interferometer. Coherent light from a laseris coupled to a fiber and then split by a bi-directional couplerinto reference and object beams. The object beam illuminates thesurface. The light scattered from the object is collected by a lensand imaged on the CCD, where it is mixed coherently by the beamsplitter with the reference beam.

ESPI Applications

Conventional ESPI techniques revealdefects in layered structures. Figure 6a shows an ESPIinterferogram on a wooden panel painting with a subsurface defectbetween the pictorial film and the substrate. Figure 6b shows the phase map relative to the out-of-plane deformation due tothe non-homogeneity of the materials.

ESPI diagnostics are also very useful in the detection ofcracks. Figure 7 shows a fringe pattern revealing thepresence of a vertical crack support at the initial stage on anancient icon.

Another important application of the technique is in theexperimental testing conducted on a carbon-epoxy plate of aircraftcomponents (particular emphasis is given to fiber failure, matrixcracking, and delamination). Figure 7b shows theinterferogram on a honeycomb aircraft component with the presenceof a large disbond where the oval fringe pattern is present. Inthis case the deformation fringes have an irregular appearance,which clearly indicates the defect area.

Figures 7a and b:  (a) Fringe pattern locating avertical crack in the support of an icon. (b) A largedisbond on an aircraft component revealed by ESPI.

Advantages of the ESPI technique include:

  • Daylight operation by simply using an interference filter infront of the CCD camera
  • Reduced stability requirements
  • Compact and portable instrumentation
  • Elimination of photographic processing
  • Full digital data elaboration
  • No need for high-quality optics and a simplified opticalsetup
  • Relatively low cost.

However, there are also some limitations to ESPI. The individualspeckle must be resolved by the TV camera, requiring a relativelylarge f -number. Consequently, the speckles are clearlyvisible, thus degrading the fringe quality. Image digitalprocessing is usually required to remove noise.

A further limitation in ESPI systems is the spatial resolutionof the camera. To resolve a fringe, about 20 speckles are required.A typical CCD with 512×512 pixels can only resolve about 25 equallyspaced fringes.

Future trends in ESPI will take advantage of large area CCDs,TV-cameras with higher sensitivity and dynamic range (for example,12-bit dynamic range), and more powerful image-processingsoftware.


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.