SciELO - Scientific Electronic Library Online

 
 issue68A simplification to the fast FIR-FFT filtering technique in the DSP interpolation process for band-limited signalsCryptanalysis and improvement of Chen-Hsiang- Shih's remote user authentication scheme using smart cards author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand

Journal

Article

Indicators

Related links

  • On index processCited by Google
  • Have no similar articlesSimilars in SciELO
  • On index processSimilars in Google

Share


Revista Facultad de Ingeniería Universidad de Antioquia

Print version ISSN 0120-6230

Rev.fac.ing.univ. Antioquia  no.68 Medellín July/Sept. 2013

 

ARTÍCULO ORIGINAL

 

Non-linear joint transform correlator for micro-displacement measurement using speckle patterns

 

Medida de desplazamientos micrométricos mediante correlación de transformada conjunta no lineal de patrones de speckle

 

 

Luis G. Valdivieso González1, Jáder E. Guerrero Bermúdez2*, Eduardo Tepichín Rodríguez1

1Instituto Nacional de Astrofísica, Óptica y Electrónica (INAOE). Calle Luis Enrique Erro 1. C.P. 72840 San Andrés Cholula, Puebla, México.

2Grupo de óptica y tratamiento de señales, Escuela de Física, Universidad Industrial de Santander. A.A. 678 Bucaramanga, Colombia.

*Autor de correspondencia: teléfono: 57 + 7 + 69 94 678, fax: 57 + 7 + 63 23 477, correo electrónico: jader@uis.edu.co (J. Guerrero)

 

(Recibido el 11 de julio de 2012. Aceptado el 5 de agosto de 2013)

 

 


Abstract

This work presents a procedure for measuring displacements of tens micrometer, in one and two dimensions, based on the correlation between two speckle patterns. For this purpose, two objective speckle patterns are recorded by a digital camera -one before and one after the object has been moved- and placed in the input plane of nonlinear joint transform correlator.

Nonlinear transformation of the joint power spectrum allows a sharper correlation peak and a high signal to noise ratio. The autocorrelation peak coordinates of the first pattern are set as a reference for measuring shifts of the successive cross-correlation peak. One criterion for reliable measure is proposed. Results related with different distances sample-sensor and illuminating wavelengths at 632.8 nm and 543.5 nm are presented.

Keywords: Laser speckle, micro-displacements, nonlinear joint, transform correlator


Resumen

Se presenta un procedimiento para medir desplazamiento en el rango de micrones, en una y dos dimensiones, basado en la correlación conjunta de patrones de speckle (moteado característico de la superficie iluminada por un láser). Para este propósito, dos patrones de speckle por propagación libre, registrados por una cámara digital, uno antes y otro después del movimiento de la superficie objeto, se colocan en el plano de entrada de un correlador de transformada conjunta.

Una transformación no lineal de la densidad espectral de energía logra una alta razón señal a ruido en el plano de salida del correlador. Las coordenadas del pico de autocorrelación, del primer patrón se colocan como origen para medir los desplazamientos sucesivos del pico de correlación cruzada. Se propone un criterio para determinar el rango de medida confiable y se presentan algunos resultados para diferentes distancias superficie-sensor y longitud de onda del láser.

Palabras clave: Moteado laser, correlación de transformada conjunta no lineal, desplazamientos micrométricos


 

Introduction

Currently there is great variety of processes and systems for measuring displacements in the range of microns. These procedures are based on principles capacitive, inductive, resistive, magnetic, ultrasonic and optical [1]. The last ones are characterized by non-contact, promptness and immunity to electromagnetic interference. The procedures for optically measuring displacements are based in triangulation, time-of-flight diffraction (TOFD), moiré and interference [2, 3]. Before the widespread use of digital cameras, laser-speckle photography was the procedure, based on speckle, used to measure the magnitude and direction of micro-displacements [4]. In this paper, we propose measure displacements, in the range of 20μm - 450 μm, using joint discrete correlation of two digital speckle patterns corresponding to two positions of a diffusing surface. This involves, defining the optical system to form the speckle pattern, establish the optimal sampling rate, development of software for processing images and the calculation of the joint correlation, and finally, associate the result of the correlation with micro-displacements.

Some aspects of this procedure has been previously discussed in a brief technical note [5], here we introduce new elements, such as, the criterion to determine the reliable range of measure, measurements in two dimensions and the use of different wavelengths.

The paper is organized as follow, a brief definition of the speckle phenomenon, followed by the mathematical formalism necessary to understand the non linear joint transform correlator, the experimental, results and conclusions.

 

Laser speckle

The speckle is a granular structure resulting from the interference of multiple beams when light, highly coherent, is scattered, reflected and refracted by optically rough surfaces. These beams have random amplitudes and phases, and its superposition produces bright and dark spots due to constructive and destructive interference. The speckle exits even without forming an image of the object, in this case is called objective speckle pattern, converse when the image of the speckle pattern is formed by a lens is called subjective speckle pattern [6]. The theoretical models show that the speckle ''grain'' exhibits a geometrical structure 3D depending on the coherence properties of the light source and the characteristics of the object's surface [7]. In order to obtain an appropriate image of the speckle pattern, according the sampling theorem, the picture elements of the camera should be at least the half of the size of speckle grain. The average grain size of an objective speckle pattern is given by equation (1). [8]

In the above equation, the 1.22 factor is a result of diffraction, λ is the wavelength of the beam that illuminates the object,z is the distance between the diffusing surface and the viewing screen, and D is the diameter of the laser beam at z distance.

Speckle pattern is suitable for measuring micro-displacements because it contains coded information about the position and shape of the diffusing surface [9].

 

Joint transform correlator

The joint transform correlator (JTC), also called linear joint transform correlator, was introduced by Weaver and Goodman in 1966; it is often used in digital, optical or hybrid optical-digital object recognition and classification. It has the following advantages:

• No need synthesis of a complex filter in Fourier space.

• It is appropriate for real time applications.

• It is robust against vibration and misalignment compared with filter-based correlator (filters placed in the Fourier plane).

One of the many modifications to the classical joint transform correlator is the nonlinear transformation ofjoint power spectrum. Nonlinear joint transform correlator exhibits sharp peaks in the exit plane decreasing the uncertainty in locating the target of interest [10]. But, JTC and nonlinear JTC present some disadvantages, such as:

• Noticeable decrease of actual output bandwidth-space.

• Presence of wide side lobes that degrade the signal to noise ratio and increase uncertainty in the location of peak correlation.

• Its performance is affected strongly with rotations of the target.

The input plane of JTC, u1 (x1, y1), consists of a reference h1 (x1, y1), in this case one speckle pattern, placed joint the input scene g1 (x1, y1), other speckle pattern corresponding to displacement of the sample. This can be written according (2),

The equation (3) depicts the intensity of Fourier transform of the correlator input plane, when speckle patterns h (x1, y1) and g (x1, y1) are located in Y/2 and - Y/2, respectively.

In equation above, I(x2, y2) is the classical or lineal joint power spectrum (JPS); H(cx2, cy2) and G(cx2, cy2) are the Fourier transforms, scaled by a c factor, of the reference h(x1, y1) and the input scene g(x1, y1) respectively.

Amplitude distribution of the output plane of linear JTC, u(x3, y3), is the Fourier transform of I(x2, y2) and is given by expression (4),

First and second terms in equation (4) are the direct component of the correlation. The last two terms are the cross-correlations between g and h functions, centered in (0, Y) and (0, -Y) positions in the output plane of correlator. The symbols, and δ are the convolution and delta function respectively.

On the other hand, in nonlinear JTC joint spectral density can be transformed according to equation (5),

Where VT is a threshold value, which could be chosen to maximize the intensity of autocorrelation peak and reduce the height of the peaks of higher order.

This joint spectral density, in equation (5) can be seen as a signal coded by pulses with width d, and position, Δ.

An expansion in a series of harmonic terms of JPS nonlinear leads to (6),

In equation (6), L is the period of the JPS along y2 axis. Maximizing the ratio between the first­order autocorrelation peak respects to the direct component, require defining a threshold, VT , which is the same value of the direct component of the JPS, and is described in (7),

Finally, the output of the correlator u3 (x3, y3) is proportional to the Fourier transform of the cosine term in equation (4). With k= 1, and VT given by equation 7, the correlation output result in the expression (8),

Equation 8 shows that the correlation consists of two delta functions located at positions (0,Y) and (0,-Y) at the output plane.

 

Experimental

Coherent light beam of a He-Ne laser was used to illuminate the surface of the sample, located on a micro-displacement station (x, y), as shown in figure 1.

The diffuse component of light, reflected from the sample, is recorded by the sensor of a monochrome charge-coupled device (CCD) camera at z distance. The origin of coordinate system for measurements was established at the position of the autocorrelation peak of the speckle pattern (zero displacement of the sample).

To obtain a calibration curve, theoretical micro-displacements were made by moving the station with the sample using a step by step motor at intervals of 20-μm, after each displacement, one objective speckle pattern was recorded and after placed in the input plane of nonlinear joint transform correlator. Cross-correlation was calculated, with the aim to obtain the respective shift of the first order cross-correlation peak and relate with the displacement of the sample. The experiment was done first using He-Ne laser at wavelength and 632.8 nm and after 543.5 nm. Multiple sensor-sample distances were considered keeping the wavelength of the laser constant.

Data processing involved routines of images treatment and correlation on moderately-fast computers systems, which made possible the almost real-time measurements of the motion of speckle patterns. The software used was Matlab 7.0. Roughly ten correlations per second were obtained; this quantity result to be small compared with the potential of optical correlation that is estimated typically in 400 correlations per second; but this setup could compete with hybrid optical-digital setups made with current special­purpose processors that handle typically 40 correlations per second [11].

 

Results

Behavior of cross-correlation intensity in arbitrary units (A.U) against displacements in microns (μrn), see figure 2(a) and (b), evidence that measure is reliable until 400 (μm) for horizontal displacement and 200 μm for vertical displacement, respectively; this is corroborated in figure 2(c) and (d), that show calculated displacement versus theoretical displacement.

The graphics suggest a criterion for determining the reliable range of measure, based on the comparison of the maximum of intensity of the first order cross-correlation and the intensity of the peak in the origin (which coincides with the coordinates of the autocorrelation peak). When these intensities are equal or the autocorrelation peak is higher than the first order cross-correlation peak, false measurement values for micro- displacements arise.

Note that, the results are repeatable; this can be appreciated in the symmetric error bars of unit standard deviation of measures. The error increases when the displacements exceed the criterion mention above. Moreover, when the displacement is under the criterion, the resolution is fixed by digital camera (in this case 16-μm. Finally, rectangular shape of imaging sensor determines a different range in the horizontal and vertical measure. Finally, rectangular shape of imaging sensor determines a different range in the horizontal and vertical measure.

On other hand, table 1 (see column 'Max. range') presents reliability range for micro-displacement measure using different wavelengths and different sample-sensor distance [12]. The best result is obtained for distance sample-sensor 7 cm, when speckle pattern is produced using a laser beam with wavelength of 543.5 nm and 6 cm using a laser beam with wavelength of 632.8 nm which is according with the optimal sampling criteria. In all cases, if the displacement is under the condition of ('Max range') the resolution is fixed by digital camera.

Finally, tracking two-dimensional of the sample can also be made following this setup. In figure 3, a circular path of 225-µm of radius is shown. The trajectory is sampled at interval of 30°.

Using this technique, there is no ambiguity in determining the sense of the micro-displacements, thus overcoming the double-exposure speckle photography.

 

Acknowledgements

J. Guerrero would like to say thanks to VIE-UIS by financial support to this work.

 

Conclusions

The results show that the nonlinear joint transform correlator is adequate for the measurement of micro-displacement in the range of tens of micrometers. It is possible to establish a reliable range for measure, by comparing the intensities of autocorrelation with cross-correlation peak. The ability of the nonlinear JTC to improve the ambiguity problem of the sense of displacement enables the plotting of 2D-trajectories. The range and resolution of measure is related with the laser wavelength, distance sample-camera and the pixel size of the sensor.

We hope that the increased capabilities of all­digital and hybrid optical-digital systems reduce computation time, potentially increasing the use of the proposed technique in an industrial environment, outside laboratory conditions.

 

References

1. R. Leach. Fundamental principles of engineering nanometrology. 1st ed. Ed. Elsevier Inc. Oxford, England. 2010. pp. 85-111.         [ Links ]

2. G. Berkovic, E. Shafir. ''Optical methods for distance and displacement measurements''. Adv. Optics and Photonics. Vol. 4. 2012. pp. 441-471.         [ Links ]

3. J. Duarte, F. Fernández, M. Moreno, J. Almerich, M. Andrés. ''Metrología por fibra óptica para la detección de pequeños desplazamientos''. Rev. Fac. Ing. Univ. Antioquia. No. 56. 2010. pp. 151-159.         [ Links ]

4. H. Tiziani, G. Pedrini. ''From speckle pattern photography to digital holographic interferometry (invited)''. Appl. Opt. Vol. 52. 2013. pp. 30-44.         [ Links ]

5. J. Guerrero, C. Torres, O. Gualdron, Y. Torres, L. Salazar. ''Micro-displacement by Means of Optoelectronic Nonlinear Joint Transform Correlation of Speckle Pattern''. Appl. Opt. Vol. 37. 1998. pp. 8153-8154.         [ Links ]

6. N. Gaggioli, J. Pomarico. ''The speckle phenomenon''. H. Rabal, R. Braga (Jr) (editors). Dynamic Laser Speckle and Applications. 1st ed. Ed. CRC Press, Taylor & Francis Group, LLC. Boca Raton, FL. USA. 2009. pp. 1-19        [ Links ]

7. T. Yoshimura. ''Statistical properties of dynamic speckles''. J. Opt. Soc. Am. A.Vol. 3. 1986. pp. 1032-1054.         [ Links ]

8. Oulamara. ''Subpixel speckle displacement measurement using a digital processing technique''. Journal of Modern Optics. Vol. 35. 1988. pp. 1201-1211.         [ Links ]

9. C. Gorecki, G. Tribillon. ''Speckle displacement analysis by phase correlation using a SLM-based processor''. Journal of Modern Optics. Vol. 40. 1993. pp. 973-978.         [ Links ]

10. B. Javidi. ''Nonlinear Joint Transform Correlators'' B. Javidi (editor). Real Time Optical Information Processing. 1st ed. Ed. Academic Press Inc. San Diego, CA, USA. 1994. pp. 115-183.         [ Links ]

11. D. Psaltis, M. Neifeld, A. Yamamura, S. Kobayashi. ''Optical memory disks in optical information processing''. Appl. Opt. Vol. 29. 1990. pp. 2038-2057.         [ Links ]

12. L. Valdivieso. Medida de micro-desplazamientos mediante correlación conjunta de patrones speckle. Undergraduate project, Universidad Industrial de Santander. Bucaramanga, Colombia. 2005. pp.15-69.         [ Links ]