Proceedings of Topical Meeting on Optoinformatics (pdf-format, 1.21 ...

14 OPTOINFORMATICS’05
DETERMINATION OF SPECKLE DISPLACEMENT BY HYBRID
OPTICAL-DIGITAL SPECKLE CORRELATOR
Sakharuk O. M., Fityo N. V., Muravsky L. I., Yezhov P. V.*
Karpenko Physico-Mechanical Institute <strong>of</strong> NAS <strong>of</strong> Ukraine, Lviv, Ukraine
*Institute <strong>of</strong> Physics <strong>of</strong> NAS <strong>of</strong> Ukraine, Kiev, Ukraine
The comparative analysis <strong>of</strong> the hybrid optical-digital speckle correlation
technique and digital speckle correlation technique is carried out. Description
<strong>of</strong> the hybrid optical-digital speckle correlator (ODSC) with first digital stage
and second optical stage is represented. The systematic and random errors <strong>of</strong>
speckle pattern’s displacements obtained by ODSC are analyzed.
Recently, methods <strong>of</strong> digital speckle correlation have been used for variety <strong>of</strong> nondestructive
testing problem solutions [1] . However, the traditional digital speckle correlation
(DSC) techniques doesn’t cover advantages <strong>of</strong> nonlinear transformation and spatial
filtration <strong>of</strong> image spectrum for improvement <strong>of</strong> correlator performance, namely decrease
noise and narrow correlation peak, which lead to exact definition <strong>of</strong> peak position and thus
displacement <strong>of</strong> an image. Only Chen et al. [2] have used the Kumar-Hasselbrook filter [3] to
improve DSC performance. Later, Muravsky et al. [4] have proposed optical speckledisplacement
correlation technique based on joint transform correlator architecture for
study <strong>of</strong> in-plane speckle displacements, where nonlinear transformation (median and
subset median binarization) and filtration (fringe adjusted filter) <strong>of</strong> a joint power spectrum
(JPS) <strong>of</strong> two input images were used.
We have created hybrid optical-digital speckle correlator with digital first stage and
optical second stage that possess advantages <strong>of</strong> nonlinear spatial transformation and
filtration. This correlator is combined with a tensile-testing machine, which apply the
external loading to studied specimen <strong>of</strong> structural material. A strainless surface S 1 and a
strained surface S 2 <strong>of</strong> the specimen are illuminated by a light source. The CMOS-camera
captures the speckle patterns <strong>of</strong> these surfaces and enters these patterns into the PC (ODSC
first stage). To study systematic and random errors, we have used computer-generated
speckle patterns r and g. The PC divide speckle patterns r and g into the equal quantity <strong>of</strong>
identical subimages r mn , and g m , n and produces the joint spectrum (JS) <strong>of</strong> each
corresponding pair <strong>of</strong> these subimages R * m,n S m,n . Then PC performs JS interpolation and
nonlinear transformation to raise accuracy definition <strong>of</strong> correlation peak position at output
<strong>of</strong> ODSC. Further, the transformed JS is inserted into an electrically addressed spatial light
modulator (EASLM), which can be treated as the second stage's input. The second stage is
an optical Fourier processor OFP containing a laser diode and Fourier lens. The
transformed JS is recorded on the EASLM and is read by a laser beam. A correlation
response is produced on the ODSC output. The correlation peak is detected by a sensor
array <strong>of</strong> a camera and its coordinates are defined by using PC. Usage <strong>of</strong> time-consuming
algorithm <strong>of</strong> subpixel resolution for peak determination can be omitted in given system by
recording <strong>of</strong> correlation peak with the whole array <strong>of</strong> camera.