Coherence probe microscopy represents a contemporary method for non-destructive investigation of micro-materials providing depth resolved information about the sub-surface domain. We present that Fourier plane filtering can be integrated in such a low-coherence interferometric imaging system and demonstrate how image contrast can be modified and enhanced, similar to microscopy but now additionally in a depth-resolved way. Thereby we focus on weakly scattering technical multi-layer structures.
Low-Coherence Interferometry in Material Research
Non-destructive testing (NDT) techniques are favorite methods in diverse fields of material research applications due to their non-invasive investigation character. In non-destructive probing, computer tomography (CT) and ultrasound (US) imaging are well established. However, they are restricted in their application by the hazards of X-ray radiation, by obeying typical resolution limits in relation to the sample dimensions, or they require an additional coupling medium.
For contactless investigation of semi-transparent stratified or scattering materials, low-coherence interferometry (LCI) - working typically in the near infrared wavelength range (NIR) - can provide an alternative to CT and US methods. In particular, the characterization of the internal structure of semi-transparent, scattering specimens can be achieved by optical coherence tomography (OCT) (in case of standard numerical aperture (NA) objectives) or coherence probe microscopy (OCM) (in case of high NA objectives). Originally, OCT was introduced for medical diagnostics . Recently, OCT has been successfully applied to technical material characterization .
The combination of OCT with elements of microscopic imaging and area cameras results in so-called full-field OCT (FF-OCT) and coherent probe microscopy (CPM) [3 ,4], whereas a point-wise detection scheme is standard in OCT. Similar to the development in conventional OCT, FF-OCT or CPM have found their application in technical material inspection and characterization.
Lateral and depth-resolved information about structures in the micron size range - e.g. inclusions, filler particles, defects or other imperfections inside the material - can be obtained. In figure 1, a fiber-reinforced polymer structure and a multilayer foil specimen investigated by CPM imaging are shown. These typical scattering or stratified examples of technical nature illustrate the abilities of CPM for non-destructive micro-material inspection.
CPM and Fourier Plane Filtering
The imaging setup is based on a low coherence interferometer of Mach-Zehnder configuration as sketched in a simplified way in figure 2a. The imaging of the sample under investigation is performed in the reflection regime. As a low-coherence illumination source we either use a superluminescence diode (from SuperLum) or a supercontinuum source (LEUKOS SM-30), emitting light in the near infraread range (center wavelength of 840 nm or 825 nm, spectral bandwidth of 50 nm or 350 nm, respectively), providing a lateral and axial resolution of up to 1 micrometer in the material. A piezo transducer then alters the phase relationship between sample and reference wave field by shifting the reference mirror in fractions of the wavelength. First, multiple phase shifted images are recorded as raw data encoding information about the embedded structures in the amplitude modulation of the local fringes.
After demodulation according to conventional phase stepping or quadrature approaches, the resulting amplitude image (CPM image) represents a horizontal section of the sample (en-face images). By coherently sensing the intensity of the backscattered light at different depth positions of the sample, we finally obtain a 3D image stack.
Since CPM no longer relies on point-wise raster scanning, techniques well-established in microscopy can be included in the low-coherence interferometric full-field setup. We incorporate a Fourier plane filtering (FPF) unit realized in 4f configuration into the sample arm. Correspondingly, an optical compensation unit balancing the optical path length and dispersion is brought into the reference arm, as sketched in figure 2b. We achieve flexible FPF by using a spatial light modulator (Holoeye Photonics AG, Pluto-NIR) being addressed with different filter functions (emulating e.g. dark field, phase contrast, Schlieren contrast, or spiral phase contrast, amongst others) .
Contrast Modification and Image Enhancement
Due to the coherent imaging nature of CPM, images can be affected by speckle noise and specular artifacts reducing the obtainable signal-to-noise ratio. In this case, post-hoc image processing with advanced mathematical algorithms provides enhancement of the quality of the image. For highly scattering specimens, we follow this procedure (fig. 1c) to increase the image information content e.g. information about internal fiber structures and composites .
Alternatively, for layered, almost transparent structures within weakly scattering materials providing only low contrast in CPM, we can directly modify the appearance of these structures by means of FPF during imaging. This allows us to emphasize relevant image features as represented by edges or interfaces.
Inhomogeneous varnish layers on a glass slide (applied to both, surface and backside of the slide) act as a test model. We can discern between features of the front and back side in case of CPM imaging, in contrast to the conventional microscopic image, where the structures are imaged in a simultaneous way, only slightly distinguishable by defocusing, as shown in figure 3a-c.
The conventional CPM setup operates in bright field (BF) mode (i.e. without FPF). Since the achieved appearance/contrast may be insufficient, especially for structures being almost phase objects, we now use the FPF mode of the setup. In comparison to microscopy the contrast can be modified and selectively improved at the depth position corresponding to the coherence gate (see figure 3d-f). The altered appearance of the investigated structures resulting from different filter functions is demonstrated for the layered test structure. If working in the dark field, Schlieren contrast, or spiral phase contrast mode of the interferometric setup, the object boundaries and interfaces become much more visible compared to BF mode (depicted in figure 4a-c), a fact well-known for 2D microscopy , but here realized for low-coherence interferometric imaging.
Furthermore, using either the full spiral phase filter (SPF) or masking the SPF by angular cone-like regions (applying a complex-valued filter function) enables us to enhance edges either in an isotropic or anisotropic way , as illustrated in figure 4d-f. It is worth noting that a certain analogon to optical full and cone-like spiral phase filtering can be found in analytic and monogenic signal and wavelet approaches in image processing. There, special methods used for the detection of salient points, like edges or corners, or for the estimation of local orientation, involving Riesz and Hilbert transforms in a mathematical way, should be mentioned [9,10].
Additionally, variable phase stepping in phase contrast imaging can easily be realized with the SLM by introducing multiple phase shifts between diffracted and undiffracted wave components. This permits us to go towards quantitative phase low coherence imaging, as illustrated in figure 5.
In our CPM modality, we have successfully combined two approaches: depth resolved imaging (based on the low-coherence interference ability of CPM) and contrast modification by FPF. We have demonstrated that, similar to the well known contrast modification in 2D microscopy, in low-coherence interferometry a contrast modification can also be achieved, in particular for layered structures and interfaces containing features such as edges or borderlines. We expect interest in these techniques in the field of technical polymer research and see potential for imaging in micro-fluidics and for applications of biological background.
The financial support by the Federal Ministry of Economy, Family and Youth and the National Foundation for Research, Technology and Development is gratefully acknowledged. We thank Prof. Monika Ritsch-Marte and her team for continuous support.
 Huang D. et al.: Science 254(5035), 1178-1181 (1991)
 Stifter D.: Appl. Phys. B 88, 337-57 (2007)
 Dubois A. et al.: Appl. Opt. 41, 805-812 (2002)
 Sheppard C.J.R. et al.: Appl. Opt. 43, 1493-1502 (2004)
 Schausberger S.E. et al.: Opt. Lett. 35, 4154-4156 (2010)
 Schlager V. et al.: Springer LNCS 6688, 424-434, (2011)
 Fürhapter S. et al.: Opt. Express 13, 689-694 (2005)
 Heise B. et al.: Proc. SPIE 8335, 83350G (2012)
 Koethe U. and Felsberg M.: Springer LNCS 3459, 179-191, (2005)
 Unser M. et al.: IEEE Trans. Image Process 18, 2402-2418 (2009)