Spectral Low Coherence Interferometry: A Complete Analysis of the Detection System and the Signal Processing

Low coherence interferometry (LCI) is an optical technique in which light is used as an instrument to obtain high resolution optical images in a great diversity of materials. It is possible to find in the literature, a variety of configurations with different names, based on the same principle: low coherence reflectometry (OLCR), optical coherence tomography (OCT), white light interferometry (WLI), are only some examples of this. The main idea of this technique is to measure the echo time delay of backscattered light in the sample through the characterization of the interference intensity obtained when the light coming from the sample and the light reflected in a reference surface overlap. When a low coherence source is used, the interference signal is temporally and spatially localized, so it is possible to use this property to obtain distance values or parameters related to the time of flight of the light, reflected in different sections of a sample.


Introduction
Low coherence interferometry (LCI) is an optical technique in which light is used as an instrument to obtain high resolution optical images in a great diversity of materials. It is possible to find in the literature, a variety of configurations with different names, based on the same principle: low coherence reflectometry (OLCR), optical coherence tomography (OCT), white light interferometry (WLI), are only some examples of this. The main idea of this technique is to measure the echo time delay of backscattered light in the sample through the characterization of the interference intensity obtained when the light coming from the sample and the light reflected in a reference surface overlap. When a low coherence source is used, the interference signal is temporally and spatially localized, so it is possible to use this property to obtain distance values or parameters related to the time of flight of the light, reflected in different sections of a sample.
LCI has been proposed and studied since the beginning of the optical science. However, for the last 20 years, there has been a dramatically increase in its applications, mainly due to the development of optical coherence tomography and the evolution of new light sources. The first works of in-surface and optical material characterization was reported in 1960. In 80´s some applications in fibre optics characterization (Takada et al, 1987), topography surfaces and internal structures in transparent media (Youngquist et al, 1987) were proposed with a LCI set-up. It is generally accepted that first biological application was reported by (Fercher, 1988). After these first works, optical coherent tomography (OCT) became a powerful technique for medical diagnosis, (Huang, 1991), in which images of the human retina and coronary artery were obtained. Since then, OCT has evolved, and nowadays it is a wellestablished technique for ophthalmic diagnosis and other biological tissues (Brezinski, 2006). New developments in light sources, fibre optic elements, detectors and processing techniques allow a dramatic increase in resolution and speed of image acquisition. These advances transform this technique in a powerful three-dimensional visualization method which has a wide diversity of applications. Over the last ten years its growth has been explosive; proof of this is the increasing number of publications, patents and companies involved in this subject (www.octnews.org). There is no doubt that this evolution has been the result of the numerous successful OCT applications in the medical field, especially in ophthalmology (Drexler et al, 2008). Over the last years its application in material characterization and non-destructive techniques is another field that is growing fast (Goode, 2009). Micro structures and MEMS characterization, surface topography, structural parameters in semitransparent materials, analysis and visualization of structural vibrations, are only a few examples (Bruce et al, 1991), (Wiesauer et al, 2005), (Wiesner et al, 2010).
Nowadays, there are several companies that offer commercial systems based on this tecnique. Polytec ® , Thorlabs ® , Carl Zeiss ® are only some examples. Figure 1 show an image obtained with the Stratus from Carl Zeiss ® . This equipment is capable of exploring the eye at a speed of several thousand lines per second. Another example is shown in figure 2; TSM-1200 TopMap ® µ.Lab from Polytec ® that can be used to acquire high-resolution topographical maps of functional surfaces and microstructures. High-speed 3D OCT imaging can provide comprehensive data that combines the advantages of optical coherence tomography and microscopy in a single system. Shown below are some 3D image data sets of two samples; in figures 4 it is shown how this technique can obtain the surface topography of a screw. b a www.intechopen.com

Basis of low coherence Interferometry
Interference can be considered as the wave pattern that is obtained after the superposition of two or more waves. In low coherence interferometry, superposition is obtained with a broadband light source, and the wave pattern in this situation is considerably different from that obtained with a standard monochromatic source. (Born, M. & Wolf, E., 1999).
In interferometry techniques the interference signal is usually obtained from the superposition between light backscattered from a reference arm and a sample arm as it is shown in figure 5 below. The total intensity at the detector at a time t, is obtained as the superposition of e r (t) = Re(Er(t)) and e s (t) = Re(E s (t+τ)), the electric field from the reference and from the sample respectively. Re(E(t)), indicates the real part of the complex expression of the electric field, E(t). τ is the difference in time of flight of light in each arm and it is a consequence of the different optical path length. The symbol < > indicates temporal average, K is a constant factor.
The reflectivity in the interface at the end of the reference and the sample arm are considered by the coefficient R, which is assumed constant. If Io is the intensity at the output of the light source, the intensity of each arm after the reflections is: I r = R r I o and I s = R s I o .
Then, equation 1 can be written: In the last term in equation 2, usually named the interference term, the complex coherence function (Goodman, 1984); that is the correlation of the electric field. This function is closely related to the concept of coherence time -T -a measure of the delay between two beams which is necessary to blur the interference term. A useful way to define this concept is (Goodman, 1984): It is possible to illustrate this concept -close to the basic idea of low coherence interferometry-considering the total intensity I as a function of τ, the time delay. To do this, we perform an axial scan of the reference sample. In the simplified interferometer scheme of figure 5 this can be done with a movement of the reference surface with a constant velocity (axis z figure 5).
When the light source is monochromatic, the total intensity shows the same variation (figure 6-a). Using a low coherence source, interference is only observed when the path length is within the coherence length of the light source ( figure 6-b).

Experimental configurations
We will show a generic system of low coherence interferometry with a brief description of the experimental set-up and the function of each component. The general configuration can be separated in four well differentiated sections: the broadband light source, the interferometer, the sample with its holder and the detection system. Figure 7 shows the basic set-up used in most of the low coherence applications. The initial system is the broadband light source [1], that is a critical factor since an appropriate selection in its characteristics is crucial in the final quality of the images obtained (Drexler et al, 2008). Nowadays there is an important offer of broadband sources with different center wavelengths, power, and beam quality (stability, noise, single transverse mode, etc.).
Diode systems are probably the most utilized light sources, as the superluminiscent LED`s (SLD). They have the advantage to be easy to hand and operate; also, their spatial dimension is reduced, their price is usually lower than other systems, and they are appropriated for nolaboratory applications. Amplified spontaneous emission (ASE) sources are an interesting alternative as they offer a broad bandwidth in the infrared region. Finally there is a group based on ultra-short laser pulses and supercontinuum light sources which is an attractive option as it offers high power, broad bandwidth and good quality in spatial mode. These systems are expensive comparatively.
The output beam from the source is directed to the interferometer. There is a great variety of configurations proposed in the literature, but in most of these works a Michelson type is used (Schmitt et al, 1999). A typical configuration is shown in figure 7. The output beam is split in the beamsplitter [4]. One of the beams goes to the reference surface [2]-usually a mirror-and the other beam goes to the sample [3]. After the reflection on each surface, both beams are sent to the detector [5]. The superposition generates the interference signal.
With this signal it is obtained a measurement of the OPD between both arms. In most cases the light beam is focused on the sample so that the measurement process is performed point to point on the desired region. For opaque samples, the light is reflected only on the surface, so a topography measurement is obtained. For transparent or semi-transparent media, light is reflected from sub-surface structures in the sample, so a tomography measurement is performed.
Usually the sample-holder or the light beam can be displaced in a 2-D lateral movement (axis x and y in the figure 7). This way the sample can be inspected in the region of interest.
Although there are many configurations proposed based on the general configuration described before, it is possible to make a division into three main groups, each one with its particular characteristics. They are commonly known as: 1. Time domain low coherence interferometry (TDLCI) 2. Spectral domain low coherence interferometry (SDLCI) 3. Sweep source domain low coherence interferometry (SSDLCI) In TDLCI mode the total interference is obtained with a one element detector( [5] in figure 7), usually a photodiode, and the total intensity is registered while the optical path of the reference arm changes from a maximum to a minimum value that is predefined for each setup (displacement in the z direction - figure 7). While the reference arm is moving the sample is maintained in the same position (Drexler et al, 2008).
In SDLCI mode the detection [5] is performed with a spectrometer. In this scheme the interference signal is obtained with the superposition of the spectral intensity of both arms. (Drexler et al, 2008).

Fig. 7. A typical low coherence interferometry experimental set-up.
Based on the same idea, in the SSDLCI configuration, a tuneable laser and a one element photodetector are used. The light source is swept in wavelength as a function of time, so the spectral components are not encoded by spatial separation, they are encoded in time. After a complete swept we obtain the spectrum one by one wavelength (Choma et al 2003), and consequently the interference signal.

Spectral domain low coherence interferometry 2.1 Introduction
Spectral domain low coherence interferometry (SDLCI) -also known as Fourier domain low coherence interferometry (FDLCI)-is, as mentioned before, one of the configurations commonly used in low coherence interferometry. The first result using this technique (Fercher et al, 1995) was for ophthalmological measurements and some years later (Hasler et al, 1998) in dermatological applications. After these initial works, it began to be shown as a competitive technique with the time domain method (TD-OCT), a well established method for optical coherence tomography (OCT) applications at that time. Since then, SDLCI has shown several advantages which have favoured its development and the high level of acceptance that it has nowadays.
These advantages can be summarized in the lack of need for a fast mechanical scanning mechanism (Drexler et al, 2008) that brings the simplicity of no moving parts in the reference arm of the interferometer, and the superior sensitivity of the detection (Leitgeb et al, 2003). The SDLI typical configuration is shown in figure 8. The light source and the interferometer scheme follow the same characteristics described before, but the detection system is a distinct point in this technique. The main idea is that the optical path difference www.intechopen.com between the sample and the reference is obtained from the analysis of the superposition of the spectrum of the light source reflected in each arm of the interferometer. This spectrum is commonly measured with a spectrometer. Most of the detection systems employ linear arrays as a sensor element. Some works with 2-D CCD systems and individual processing of each pixel, has been already presented (Vakhtin et al, 2003).

The interference signal and the detection system
It is assumed that I o (k) encodes the power spectral dependence of the light source ([1] in figure 8), were k=2 /λ, is the wavenumber. The interference fringes surges from the superposition of the spectrum of the light coming from the two reflections. We call I r1 (k) to the intensity coming from the reference arm and I s1 (k) to the intensity coming from the sample arm . The expression for the total intensity is then: The first two terms are known as DC intensities in the literature (Drexler et al, 2008). The last term in equation 4 is the interference component, and it includes the OPD dependence that, in this simple situation, is given by the value ΔX r1s1 = x r1 -x s1 , where x r1 and x s1 are the total path length in each arm, both measured from the beamsplitter.
To simplify the expression, we use β 2 = R s1 /R r1 , the reflectivity coefficients ratio, and it is assumed that these coefficients have no dependence on the wavenumber, that is: I r1 (k) = R r1 I o (k) and I s1 (k) = R 1s I o (k), as mentioned in equation 5, so: The figure 9 shows an example of an image obtained in a 2D-CCD sensor at the end of the spectrometer (see set-up picture). The image corresponds to the total intensity described www.intechopen.com before in equation 4. In this particular example, the source is a superluminescent diode ( o = 840 nm, Δ = 20nm) and R 1s = R 1r  0.08.  In a general situation in which it is assumed that there are several reflections in the reference arm and in the sample arm, the total intensity expression can be written as:

Interference
N represents the total number of reflectors in the sample and in the reference. The sub index i or j identifies the region where the reflection is produced in the reference (r 1 , r 2 , r 3 , …) or in the sample arm (s 1 , s 2 , s 3 , …). To illustrate this point, we show a typical application in which the reference arm is a mirror (1r) and the sample is a slab in air. The slab has two interfaces (1s and 2s), an inner group index n, and a thickness d.
The first three terms are the DC terms mentioned before. The second and the third term correspond to the cross correlation between the reference and each of the sample surfaces; the forth term is the correlation component that surges from the reflections on both interfaces of the sample. (see figure 10). www.intechopen.com and x s1s2 = 2.n.d . Where xs1, xs2, xr1 and 2.nd, are the total optical path distances corresponding to each interface indicated in figure 10.
The usual way to obtain the OPD values from the total intensity data is to apply the Fourier transform. We call FI(x) to the Fourier Transform of I(k), where x is the conjugate variable of the wavenumber k.
As an example it is shown the expression for FI(x) in the situation mentioned before (equation 4), where the interference signal is obtained from only two reflections (r 1 and s 1 ). So: We denote the convolution as (  ).
According to the Wiener-Kinchin theorem (Goodman, 1984), the first term in the convolution is the coherence function Г(x) and the second term yields two delta functions located at ±Δx r1s1 (Papoulis, 1962).
Therefore, the last expression can be written as: Being α a constant factor. This graph shows three "interference peaks". One is centred in the origin of the x axis and the "lateral peaks", denoted as P and Q in the figures, are centred in ± Δx r1s1 .
www.intechopen.com Fig. 11. a) and b) shows the total interference signal as a function of wavenumber k, for two different relations. c) and d) shows its corresponding Fourier transform as a function of x (the conjugate variable of k) . Figure 12 shows an example of how the position of the interference peak changes according to the increment of OPD between both arms. The peaks in order zero no contain OPD information. Fig. 12. Interference peak vs OPD increments. www.intechopen.com

Signal detection and parameter limits
In an ideal situation, the intensity function I(k) would be obtained with a "perfect" spectrometer. In a real situation, it is important to evaluate the spectrometer parameters in order to optimize the measurements and evaluate possible artifacts, errors or limitations introduced for the "real" set-up. We assume a detector model based on a grating spectrometer with a linear array sensor. There are some important issues to be considered in this evaluation, more details can be found in (Hu et al, 2007;Jeon et al, 2011).
a. The spectrometer is used for separating the different wavelengths of the light source spectrum and to focus each of them on the sensor that is considered as a linear array of N p pixels. After its calibration it is possible to define a k axis in the array direction. The spectral range of the spectrometer (k a) ) is given by the interval [k 1 , k Np ] where k 1 is the value assigned to the first pixel and k Np to the last one. We also define the "Range" function as follows: () b. The dimensions of the spot of each wavelength focused on the sensor is taken into account by the point spread function (Psf(k)) (Hu et al, 2007). The width, the shape and the position of the center of this spot depends on the spectrometer characteristics such as the entrance slit width, diffraction grating and focal length and eventually on the spatial profile of the light source in the spectrometer entrance (Dorrer et al, 2000). We assume a Gaussian shape (Hu et al, 1997) where its FWHM (k psf ) is a measurement of the spot size.
The " ideal" detected signal is then modified by the convolution with this function: c. The value obtained in a particular pixel "i" is the total intensity integrated over a width (Δk pix ), which is defined by the pixel dimensions, the grating dispersion and the focus length of the spectrometer (Leitgeb et al 2003;Hu et al 2007). This average value can be written as: The convolution takes into account the process described before. A first approach to this function can be to assume a rectangular shape where its width is: Δk pix = Δk a / N p (Hu et al, 2007) (Wang et al, 2008). Here it is assumed by simplicity that the fill factor is 100%.
More complex dependence with k can be thought, considering additional effects such as crosstalk or fill factor (Jeon et al, 2011).
d. The measurement of the total intensity I(k) obtained from the detector is a discrete collection of N p values. Each value corresponds to the average intensity measured in one pixel, for this reason, it is convenient to define a sampling comb of delta functions as (Dorrer et al, 2000): Where k i is the value assigned to the pixel with index i (Wang et al,2006); (Hu et al, 2007).
The following figure shows a schematic of the spectrometer sensor and a detail of how two wavelengths (i and j) have been focused on it. The picture also shows how the "pix" and "psf" function combine to define the average value in pixel j. Fig. 13. Schematic of the spectrometer sensor array and the "psf" and "pixel" functions.
Then, the final expression for the signal intensity is: This expression is easily obtained from equation 18 and some mathematics relations (Papoulis, 1969). FI o is a constant factor that also absorbs all the pure phase terms that result from the Fourier transform. The new delta comb has a period x=2k pix , the interval available for signal processing avoiding aliasing.
The expression in brackets in equation 19, shows the Fourier transform of the "ideal." intensity and the modifications introduced by the Fourier transform of the psf and the pixel function. This modification means a sensitivity fall-off when the OPD value increase. In the figure 14 it is shown these terms for a typical situation.
As it can be appreciated from the figure 14, the shape of both functions (the sinc and psf Fourier transform) generate a decrease in the visibility amplitude of the interference peak that goes to zero when x increase. So, there are two parameters that limit the dynamic range of the technique. One is: x pix =2/k p , the first zero of the sinc function. The other is the FWHM of the Gaussian function obtained as the Fourier transform of the psf: x psf =4.ln(2)/k psf (Bajraszewski et al, 2008), (Leitgeb et al, 2003).

Signal processing
In the last section we showed some considerations in the interference image, imposed by the real characteristics of the detector. The SDLCI detection process ends with an array of N p In figure 15 a typical example is shown. The interference signal is registered in a 640-point array ( figure 15-a). Its conventional FFT is shown in figure 15 The OPD value x r1s1 in this scheme is the abscissa coordinate of the maximum point of the lateral peak in the Fourier transform curve ( figure 15-b). With the conventional Cooley-Tukey FFT algorithm the sampling interval is: Where N p and Δk pix are the parameters previously defined. In this step, spectral calibration is a critical process. First, a conversion of the measured spectrographs from -space to kspace is needed. As the spectra obtained by the spectrometer are not necessarily evenly spaced, a posterior resample to be uniformly spaced in k-axis is required. This implies a careful calibration of each pixel of the sensor. There are several works in which this point is presented (Dorrer et al, 2000;Hu et al, 2007) and different experimental and post process arrangements are proposed (Bajraszewski et all, 2008;Jeon et al 2011).
Another point that can be avoided or smoothed by signal processing is the edge effects in the Fourier transform of the Range function proposed in equation 10, to take into account the spectrometer range. To modify the shape of the rectangular function the windowing technique can be used (Oppenheim et al, 2000). With this tool it is possible to minimize the effects that result in spectral leakage in the FI(x) signal, and to increase the resolution in the OPD measurement.
There are many types of windows available in the literature. Figure 16 shows common functions (Rectangular, Bartlett, Hanning, Hamming, Blackman, Blackman-Harris and Gaussian).

c)
A spectral analysis of the different window functions is recommended to determine the best to use. In some cases it is possible to apply more than one window simultaneously.
To increment the resolution some techniques can be used (Boaz Porat, 1997). One of the most popular is known as Zero Padding (ZP) (Dorrer et al, 2000;Yun et al, 2003). In this technique the original array I[k] is extended with zeros, then, the length of the new array changes to M p > N p . As a consequence, the sampling interval can be reduced by a N p /M p factor with the disadvantages that mean computing longer vectors and more time consuming.
An alternative technique that allows reducing the sampling interval x samp in a similar way to ZP but in a localized region of the x axis is the Chirp Fourier Transform (CFT). This predefines the sampling interval and the region of interest on the x axis. Where Δθ is the sampling resolution desired, θ 0 the initial x value and K the number of additional points. Therefore, through θ 0 , Δθ and N p , you can delimit the frequency range in which you want to work. The operation number required is approximately M+N.
In figure 17, figure 17-a). To increment the resolution in a factor of 5, we used ZP and CFT. The resolution level is equal in both cases but FI ZP [x] has 3200 points and FI CFT [x] has only 81 points as showed in figure  17-b and 17-c.

Experimental results
In this section we show results obtained from an experimental configuration similar to that presented in section 2.1. The idea is to show that with a relatively simple set-up, it is possible to apply this technique in a variety of applications with interesting results. The light source is a superluminescent infrared diode, centered at 840nm, with a 20-nm bandwidth and a 5-mw output power. The interferometer is a Michelson type, in air , and the detector is a spectrometer Ocean Optics model 4000. With this configuration we get a 2-mm dynamic range and the axial resolution is lower than 10 microns. At first a calibration process is shown. As it has been mentioned before, it is necessary to link the interferometer signal with the OPD absolute values. A simple way to do this is to move the reference surface through controlled displacements (OPD) or steps, and simultaneously register the abscise coordinate of the center of the FTI peak that corresponds to each step. Figure 18-a shows the peak abscise coordinate (pixel number) versus displacement. Figure 18-b shows the peak amplitude versus the corresponding displacement. The amplitude fall-off shows the consequence of the "real parameters" of the detector explained in section 2.3. For this particular set-up the psf function is predominant.
When the OPD value is close to the limit imposed by Nyquist, aliasing effects can affect the interference signal (Dorrer et al, 2000). If the Nyquist limit is transposed, the measurement shows no-real OPD values that can be misunderstood ( figure 19).
In the next example the technique is used to obtain the profile of a metal sample. The sample is a gauge (class 2-norm ISO 3650, model M7T the C.E. Johansson Inc.); its nominal thickness is 1100 ± 0.45 m; this gauge was placed on a second gauge which was used as a reference plane, as shown in figure 20. A 4-mm length profile was obtained on the region indicated in the figure 20 (2 mm on the gauge surface and 2mm on the reference plane). The profile is obtained by measuring the axial distance in a sequence of points on the sample surface (figure 21-a). Each point is measured after a lateral displacement by steps of 50 microns. The value of gauge thickness obtained from this measurement is: 1100 ± 0.71 m. Figure 21-b shows the dispersion in each point measured, which is much lower than the coherence length of the source, except on the borders of the sample step.
In the following example an application on 3D-surface measurement is shown. These measurements are critical to the successful manufacturing of precision parts. Components and structures ranging from submillimeter to centimeter size can be found in many fields including the automotive, aerospace, semiconductor and data storage industries. www.intechopen.com The last example is to show a different configuration in the interferometer system.
With this set-up it is possible to obtain measurement of two sample's face simultaneously employing a third reference surface, which is an advantage over other techniques as the interferometric gauge block, (Decker & Pekelsky, 1997). This is very useful for thickness measurements in opaque samples or where the refraction index is unknown. In figure 23 it is shown a basic Sagnac-Michelson interferometer set-up (Morel & Torga, 2009). This is a ring interferometer type were M 1 is the reference mirror; M 2 and M 3 are the mirrors employed to send the light to each face of the sample (S in the figure 1).
BS 1 and BS 2 are two beam splitters. The light source is a superluminescent diode and the detector system ends in a spectrometer. The reference arm ends in mirror E 1 . After BS 2 , the two beams are directed to each of the faces of the sample; after their reflections in the sample face and reflection E 1 , we obtain three interference signals which let us know the sample thickness. We measure the interference signals in the detector system.
In order to identify the origin of each of the interference signals employed in the measurements, we define D r as the optical path difference (OPD) between the reference arm and the beam in the ring interferometer when the sample has been removed; D 2 and D 3 represent the OPD between the reference arm and the reflections in each of the sample faces, and D 1 is the OPD between both samples faces.  We showed that it is possible to get the sample thickness (d) with the following relations: As D 2 and D 3 are the OPD between the sample and a reference plane, a lateral displacement of the sample let us obtain the surface topography of each of the faces. From the same curves it is possible to improve the alignment of the sample with the reference plane (in our set-up the mirror E 1 ).
An example of a typical measurement obtained with this set-up is shown in figure 25. The nominal thickness of each gauge is 1.1 mm and 1.05 mm. Both are positioned so that one side has an exposed step and the other side forms a planar surface. The total lateral displacement in the experiment is about 10 mm. We called e1 to the step between the two gauges; e 3 and e 2 are the thickness of each gauge. Fig. 24. Schematic of the sample.
In figure 25.a it is shown the profile obtained from the measurement of D1. Figures 25.b and 25.c show the profiles of both faces of the sample (D2 and D3). From these results it is possible to obtain the height of the step (e1 in figure 24) and the thickness of both gauges (e3 and e2 in the figure 24). The slopes of these curves let us to obtain the relative alignment of each of the sample faces with a reference plane and the relative alignment between both gauges.
The values obtained for the gauges thickness are 1091.7 µm and 1047.46 µm, with an average dispersion of 1.32 µm, in very good agreement with the expected value.
There is a dark zone in the curves that appears when the sample presents abrupt changes in surface topography, so the light is scattered in high angles and the collected light intensity is under the detection sensitivity. The optic to focus light on the sample can be selected to minimize dark zone usually lowering the dynamical range of the interferometer.
This method enables simultaneous measurements of physical thickness and refractive group index without any prior knowledge on samples as showed in (Park et al, 2011). www.intechopen.com

Conclusions
In this chapter we present a brief introduction to the theory of low coherence interferometry and the main configurations used nowadays. Then, we focus in giving a detailed analysis of the technique "Spectral domain low coherence interferometry" that offer interesting advantages over alternative methods. We give a description of the limitations and experimental design considerations to take account in the detection of the interference intensity and in the signal processing. We also show how this knowledge is important to improve the information obtained from the images.
At the end we show some experimental results with the intention of giving examples of the enormous potential of this technique, specially in the non-destructive tests area, where we think is nowadays largely unexplored.

Acknowledgment
E. Morel and J. Torga are professors at the Universidad Tecnológica Nacional and are members of the Consejo Nacional de Investigaciones Científicas y Técnicas de la República Argentina.
Thanks to Hernán Miranda for his collaboration and special thanks to Pamela Morel for her support and images design. This book provides the most recent studies on interferometry and its applications in science and technology. It is an outline of theoretical and experimental aspects of interferometry and their applications. The book is divided in two sections. The first one is an overview of different interferometry techniques and their general applications, while the second section is devoted to more specific interferometry applications comprising from interferometry for magnetic fusion plasmas to interferometry in wireless networks. The book is an excellent reference of current interferometry applications in science and technology. It offers the opportunity to increase our knowledge about interferometry and encourage researchers in development of new applications.

How to reference
In order to correctly reference this scholarly work, feel free to copy and paste the following: Eneas N. Morel and Jorge R. Torga (2012). Spectral Low Coherence Interferometry: A Complete Analysis of the Detection System and the Signal Processing, Interferometry -Research and Applications in Science and Technology, Dr Ivan Padron (Ed.), ISBN: 978-953-51-0403-2, InTech, Available from: http://www.intechopen.com/books/interferometry-research-and-applications-in-science-andtechnology/spectral-low-coherence-interferometry-a-complete-analysis-of-the-detection-system-and-thesignal-pro