Nonlinear Dynamics and Complexity in Optical Physics:
Nonlinear Optoelectronic Image Processing

Parallel image processing and particularly detection and tracking of moving objects underpins many military and commercial applications.  At present existing image technology is neither fast enough nor sufficiently flexible to handle real-time and high-resolution data of low contrast and in cluttered scenes.  This programme undertakes research on novel nonlinear partial-differential equation based algorithms for real-time and high-resolution signal processing, and from this to develop a hybrid nonlinear opto-electronic device to undertake these tasks.  The novelty of this project is three fold: development of nonlinear PDE algorithms for real-time and high-resolution image processing, experimental implementation of such algorithms using a hybrid optical-electronic architecture, and use of electronically programmable nonlinearity with diffractive optical feedback and system integration with adaptive optics in the hybrid system.  Latest developments in optical micro-electro-mechanical systems and microelectronics make possible the full integration of the hybrid system to a compact device.  The subject of this research is currently a mainline research activity worldwide.  This programme involves QinetiQ in the UK and the Army Research Lab (ARL) in the USA.

The proposed image-processing system is a generic hybrid opto-electronic device, comprising a phase spatial light modulator (SLM) coupled with a photo-detector array through a combined optical and electronic feedback loop.  As shown in the schematic, images under investigation are introduced to the system via the SLM as phase modulations on a coherent laser beam when it reflects from the SLM.  Before they are fed back to the SLM, the phase modulations are processed, both optically and electronically according to which is best suited for a particular task.  For example, (optical) wave diffraction for image edge enhancement and (electronic) Kerr nonlinearity for phase distortion compensation.  From a theoretical point of view, the experiment can be accurately modelled by a set of coupled partial differential equations (PDE).  Our simulation results, examples of which are shown below, will therefore provide a reliable guideline to the experiment.

Figure: Spatiotemporal Transfer Function for objects of three different velocities.  q/q0 is the normalised transverse wave number.

Figure: (a) input image shows a moving object from t = 0 to t = 300; (b) a snapshot of output image after suppression of background; (c) and (d) the original and processed image of a snapshot of a video stream showing a moving car in country road.

Back to Research Reviews' Page