The authors propose a real-time recognition device using a CGH (computer-generated hologram) and an LCD (liquid crystal display). It has high speed due to optical parallel processing, and has flexibility so that the contents of the recognition can be easily changed. It is shown experimentally that the optical correlator, i.e., the subsystem of the proposed device which has the CGH and LCD, can be realized. To use the optical correlator for machine vision, however, rotation and scale invariance of objects is needed. For this purpose the authors discuss the lnr- theta coordinate transformation which results in rotation and scale invariance. It is verified by computer simulation that the proposed device can recognize objects without regard to the scale and rotation variance of objects.<<ETX>>
[1]
大津 展之.
An invariant theory of linear functionals as linear feature extractors
,
1974
.
[2]
D Casasent,et al.
Real-time deformation invariant optical pattern recognition using coordinate transformations.
,
1987,
Applied optics.
[3]
Emil Wolf,et al.
Principles of Optics: Contents
,
1999
.
[4]
W. H. Lee,et al.
Sampled fourier transform hologram generated by computer.
,
1970,
Applied optics.
[5]
A W Lohmann,et al.
Binary fraunhofer holograms, generated by computer.
,
1967,
Applied optics.