The correlator technology and concepts can be used in a wide range of application areas, which fall into either or both of two categories, namely recognition/validation/ inspection and motion tracking. It is of particular use in applications involving extremely large amounts of data processing. Below is a summary of application areas CCL is either currently involved with or has interest in. Other fields include Database Searching, Component Monitoring, CCTV Real Time Analysis, Mass Spectrometry, Automated Vision Systems.
By applying the principles of Fourier optics to solving Partial Differential Equations (PDEs), CCL is developing a technology which addresses the fundamental limitations of serial electronic processing. The technology is of particular significance to High Performance Computing (HPC) / supercomputer architectures used to solve PDEs in the field of Computational Fluid Dynamics (CFD), which is considered of particular relevance as its demands for increases in data resolution and processing speed cannot be met with the current FFT-based approach.
Using the same principles used as the basis of the correlator architectures, partial derivative operations may be performed optically with a modified 4f system, at arbitrary orders and directions, based upon the following Fourier transform property:
By combining multiple instances of the optical derivative function, flow analysis and Direct Numerical Simulation of large PDEs such as the Navier Stokes (below) that govern CFD may be performed optically, at rates far in excess of current and future electronic architectures. This has the potential to provide a step change in performance to the generation of the model flow data and also to the post analysis functions.
Current proof of concept optical derivative processor.
Following the initial success of the proof of concept, we are now currently looking to continue the development in a project with Cambridge University and secure long term development partnerships.
Coded Aperture Imaging is a process that is used in astronomy to image at wavelengths where conventional optics are not valid, such as X-rays. Based on the principles of the pinhole camera, the process uses a mask made of multiple "pinholes" to produce a convolved distribution at the camera sensor. This distribution must then undergo significant electronic post processing to reproduce the original, intended image.
CCL recently completed a phd internship project which examined the feasibility of using the principles of optical correlation to provide a real time method of performing the post processing reconstruction. The project also examined how the concept may be advanced further by replacing the fixed mask with a dynamic liquid crystal SLM - ideas which may provide significant weight and performance advantages to airborne systems.
The project was funded by and performed in collaboration with MBDA Missile Systems, the Technology Strategy Board Knowledge Transfer Network and the University of Sheffield. Download the summary here
Below is a video clip of our previous binary demonstrator system performing motion tracking and roadsign recognition.
With the NHS plans for screening all over 50’s for diseases coming into force over the next few years, the need for high data processing systems is set to become ever more apparent, making the CCL optical correlator a highly attractive technology.
It is still commonplace for teams of workers to examine high-resolution test data to pick out cancerous or abnormal cells (for example Pap smears). CCL is looking into how the correlator may be put to use to diminish the errors and missed warning signs in a variety of medical applications.
The ability of the correlator to examine and pick out matches from cluttered or partial data creates an exciting opportunity in the rapidly expanding field of biometric recognition. Full graphical comparisons of captured data are possible in a field where the norm is to look for specific features. But what happens when those features are not captured? The Holy Grail of using standard CCTV footage to pick out the terrorist from a single frame seemingly remains out of reach of current systems.
Fingerprint recognition relies upon identifying enough minutiae points from the sample to guarantee a match. However, many people do not have any such minutiae points and cannot therefore be included in normal search process.
In Facial recognition, most processes rely on calculating the distance between the eyes and nose, or on low-resolution eigenfaces. However, these rely on capturing a large proportion of the face and suffer from changes in appearance or expression.
With new 3-D mapping techniques to counter illumination effects and changes in appearance, facial recognition is finally set to come of age. CCL is looking to form collaborative partnerships and projects with leaders in the field of 3-D facial image capture – with particular focus on high-speed detection using high-resolution data and recognition using partially captured images.
Current accelerometer and gyroscope-based systems suffer from drift and calibration problems, whilst other methods involve placing unwanted emissive devices at fixed positions around the cockpit.
By using high speed images feeds from one or multiple cameras mounted on the helmet of the pilot, the shift invariance property of the CCL correlator can be used to provide a measure of how much the view of the pilot changes between frames. This data is crucial if the Helmet Mounted Display (HMD) system is to provide the pilot with the information he requires – in an environment where split second decisions can be the difference between life and death.
With the cameras set to look at the internal features of the cockpit, the system requires no other information and is totally independent of other aircraft systems. The current and previous frames from each feed are compared to provide the relative shift information, as shown below:
With the dawn of Voice over Internet (VoIP) and the ever increasing amounts of communication traffic constantly being sent around the world, standard electronic processes are no match for the huge amounts of processing required. CCL is looking into ways in which the correlator may analyse data streams for specific patterns or words, in even encrypted data.
This may also be applied to validating transmissions of digital or analogue data, such as comparing the transmitted data with that received for television signals for example.
PIV involves examining the motion of particles in a given space. This results in excessively large amounts of data being produced, which must be analysed as quickly as possible in order to produce successive motion vector distributions. Electronic systems only achieve this on small areas due to their resolution and speed limitations. The OPtical Correlator may be used to quickly analyse much larger areas and provide a true real-time solution.
Current analysis techniques, such as accelerometer-based systems, require contact with the structure under test. With the correlator no such contact is required and furthermore, areas may be examined rather than points, which limit the functionality of current systems.
Low frequency vibrations may be detected and measured using the correlator as a comparator, comparing sequential frames from a high-speed camera. High frequency tests, which involve measuring frequencies with small amplitudes, are likely to involve the use of an interferometer-based system, using the correlator to examine fringe patterns.
It is hoped that the application may then be extended to Condition Monitoring, where products may be examined for internal weaknesses by way of the fringe patterns produced when under vibration conditions.