We present a development of the method of synthesized phase objects (SPO-method) [1] for the rotation-invariant pattern recognition. For the standard method of recognition and the SPO-method, the comparison of the parameters of correlation signals for a number of amplitude objects is executed at the realization of a rotation in an optical-digital correlator with the joint Fourier transformation. It is shown that not only the invariance relative to a rotation at a realization of the joint correlation for synthesized phase objects (SP-objects) but also the main advantage of the method of SP-objects over the reference one such as the unified δ-like recognition signal with the largest possible signal-to-noise ratio independent of the type of an object are attained.
Synthesis of an axial optical vortex (OV) beam using a Gaussian beam and two ramps implemented by a spatial light modulator (SLM) is realized. Similar phase construction on an irregular thin plate receiving the OV beam is achieved. Angle dependences between two ramps and the characteristics of resulting field with respect to the observation distance of the resulting field are investigated. Intensity distributions of the beams obtained by a SLM and irregular an plate similar to those of a singular beam with a single charge are obtained.
As a result of optical experiments, it is established that the controlled variation of a recognition sensitivity at the comparison of the input and reference objects can be realized, while solving the recognition task within the method of synthesized phase objects,1 as distinct from the standard approach. We have obtained and analyzed the results of the recognition for a collection of objects with the use a joint Fourier-transform correlator. For each of the objects, we studied the character of the signal amplitude decrease depending on the growth of distortions in the object structure. We show that it is possible to obtain a family of cross-correlation curves that decrease differently within the method of synthesized phase objects, rather than a single curve. This allows one to vary the sensitivity of the correlator in the process of recognition.
The optical speckle-displacement correlator based on hybrid optical-digital joint transform correlator architecture with
digital first and optical second stage is used to determine correlation peak position with subpixel accuracy without usage
of intricate interpolation algorithms. Experimental setup for realization of the optical speckle-displacement correlation
technique was constructed on basis of a digital Fourier processor allowing joint power spectrum median and ring median
binarization and an optical Fourier processor. Speckle patterns of steel beam specimen (steel 45) with different maximum
spatial frequencies were recorded. As one of the joint transform correlator main parameters is distance between fringes at
correlator frequency plane, comparison of optical speckle-displacement correlator performance for different values of
speckle pattern maximum spatial frequency for the given joint power spectrum modulation was performed. Experimental
results have shown that the signal-to-noise ratio (SNR) increases steadily while the maximum frequency of speckle
pattern multiplied Fourier spectrum is reaching the Nyquist frequency fN. The analysis of the speckle pattern with
frequencies higher than Nyquist frequency has shown that the SNR growth is continued to the some boundary frequency
fB>fN after which the SNR is fallen sharply. Thus, the influence of aliasing on the correlator performance was studied and
the best correspondence between value of speckle pattern maximum spatial frequency and distance between fringes at
the correlator frequency plane was found.
A modified IFT-algorithm of synthesis of a kinoform is presented. In this method, two object-dependent filters
(the amplitude (A) filter and the phase (P) one) are formed in the object plane in the process of iteration,
rather than one phase filter, as usual. Both these filters participate in the formation of the phase structure of a
kinoform. The use of two degrees of freedom (A, P) in the object plane instead of one degree (P) accelerates the
convergence of the IFT-algorithm and allows one to reach significantly less values of the amplitude variance in
a reconstructed image as compared with those within the available methods. The results of optical and model
experiments confirming the efficiency of the method are given. It is found that the method is most efficient for
binary objects. The method is generalized to the case of off-axis kinoforms. A simple method to introduce the
carrier frequency into the structure of an axial kinoform is proposed. An image reconstructed by an off-axis
kinoform is free from the noises with zero and close frequencies caused by the nonideality of both the phase
mode of operation of a SLM and the effects of quantization of the registered phase.
The method of pattern recognition based on replacement of object images incoming to the correlator input
by object-dependent synthesized phase objects calculated using the iterative Fourier-transform algorithm was
developed by us earlier. In this work, we performed experimental testing the above method by using an opticaldigital
4F-correlator. Synthesized phase objects were introduced into the correlator through the spatial light
modulator LC 2002. Holographic matched filters were recorded using self-developing photopolymers PPC-488.
For two test objects, we obtained unified (δ-like) correlation signals with the signal-to-noise ratio reaching 24
dB, while the diffraction efficiency of these filters was up to 30%.
A new approach towards solving the pattern recognition problems in hybrid optical-digital 4F correlators is suggested. The method is based on that during the realization of the recognition procedure in the optical-digital correlator instead of real amplitude or amplitude-phase objects incoming on its input there is proposed to use synthesized phase (SP) objects unambiguously related to it. SP-objects represent object-dependent phase distributions calculated with application of iterative Fourier-transform (IFT) algorithm. Replacement of real objects with SP-objects allows to unify the shape (form) of correlation signals and so to formalize the procedure of choice of recognition attributes for objects of arbitrary classes. The modified recognition procedure in optical-digital correlators is presented, the influence of noises and distortions in the structure of input object upon characteristics of cross-correlation signal of corresponding SP-objects is studied. Numerical simulation results are presented.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.