Optical information processing
Digital computing technology — based on binary data and silicon based electronic processors — dominates the computer landscape, and with good reason: it is a profoundly impressive and mature technology. However there are other candidates. The ability of silicon based computing to sustain exponential Moore’s Law improvements in performance (doubling transistor count every 2 years) is being questioned, making now an exciting time to consider alternatives.
Figure 1. The Fourier transform of a positive focal length lens
Fourier optical information processing is one such alternative. It exploits the serendipitous occurrence of the Fourier transform in optical systems as demonstrated in Figure 1, and was heavily developed historically for applications such as pattern recognition based on optical correlation architectures such as the phase only matched filter (POMF) and the joint trasnsform corrtalator (JTC).
Figure 2. Basic principle of the optical derivative engine. Lower right shows the filters required for different derivative functions.
There is potential for a revival of this technology, re-targeted as a ‘co-processor,’ to deliver a step change in performance for certain calculations. Specifically, it is proposed to create two different coprocessors: one to perform a 2D Fourier transform with a novel algorithm to allow direct phase determination; and one two differentiate a 2D function. The basic operating principle of this all-optical derivative processor or engine is shown in Figure 2.
Optalysys Optical Computing - Explained by Professor Heinz Wolff
Optical pattern recognition (comparators)
The joint transform correlator (JTC) is one of two main optical image processing architectures which provide us with a highly effective way of comparing images in a wide range of applications. Traditionally an optical correlator is used to compare an unknown input scene with a pre-captured reference image library, to detect if the reference occurs within the input. There is a new class of application for the JTC where they are used as image comparators, not having a known reference image, rather frames from a video sequence form both the input and reference. The JTC input plane is formed by combining the current frame with the previous frame in a video sequence and if the frames match, then there will be a correlation peak as shown in Figure 4. If the objects move then the peaks will move (tracking) and if something has changed dramatically in the scene, then the correlation between the two frames is lost.
Figure 3. The 1/f JTC architecture in the lab
The JTC can be built using a single liquid crystal over silicon spatial light modulator (SLM), a Fourier transform lens and a digital CMOS camera. The system is used over 2 passes to create the final correlation output plane, with the first pass creating the joint power spectrum which is them nonlinearly processed before being displayed on the SLM for the 2nd pass. This processing gives us a lot of flexibility in the way in which the comparator works, allowing its function and sensitivity to be tuned and optimised. This includes adding invariances such as rotation, scale and occlusion.
Figure 4. The JTC used as a frame by frame comparator.
This forms the basis of a very powerful application for the JTC in Defence and Security as well as many other machine vision problems such as industrial inspection or production line monitoring. Any change in the scene can be recorded and with the inherent shift invariance property of the correlator, any movement of the objects in the scene can also be detected.
Figure 5. The 1/f JTC from Cambridge Correlators (now called Optalysys)
Further reading on the joint transform correlator/comparator:
Download a JTC presentation (pdf)
Related journal patents and publications:
Optical processing, Patent application number WO2008110779, US8610839, filed 10th March 2008.
Optical correlator, UK patent application number WO2004029746, 9726386.7, filed 12th December 1997.
Non-display applications and the next generation
of liquid crystal over silicon technology
of a binary filter by direct binary search algorithm for rotational
comparator based on FLC over silicon technology
comparator based on an FLC over Silicon SLM
optical correlator based helmet-tracking system
tracker based on a compact optical correlator
the discrimination capabilities of the joint transform correlator for
a product validation application
phase-only 1/f joint transform correlator using an FLC SLM
recogniser based on FLC over silicon technology
binary phase-only 1/f joint transform correlator using an FLC SLM
binary phase-only 1/f joint transform correlator
optical correlator using silicon backplane FLC SLMs for fingerprint
invariant optical correlators using ferroelectric liquid crystal spatial
binary holograms using an FLC SLM
invariant binary phase-only matched filter using a ferroelectric liquid
crystal spatial light modulator