Correlation and image recognition with surface-scattered light

Image recognition by use of coherent optical processors and light diffusely scattered from the surface of an optically rough object is reported. A theoretical description is presented and shows that the image speckles are carriers for the Fourier spectra of the object at the matched spatial-filter plane. Experimental results of optical autocorrelation and cross correlation are given. The change in the intensity of the correlation peak that arises from the translation and the rotation of objects and from the lateral and axial movements of the matched filter are examined. The system is shown to be tolerant to misalignments in the positions of the object and matched filter. It is also shown that, when diffuse light is input into the coherent optical processor, the position of the Fourier plane is no longer precisely defined and spatial multiplexing would be possible.