Computational imaging for surveillance systems

The aim of the proposed PhD project is to develop and apply computational imaging techniques to provide improved Size, Weight Power and Cost (SWaP-C) and new performance capabilities for military imaging systems. In traditional imaging systems the cost of lenses, manufactured from high-cost semiconductors such as germanium contributes significantly to overall system weight and cost. Several examples of how SWAP-C can be improved by use of computational imaging have been identified during previous research conducted by Harvey's group in Glasgow and Qioptiq. Two approaches are: (a) the use of wavefront coding, a technique involving optical encoding during image formation followed by computational decoding, to reduce the sensitivity of image generation to optical aberrations enabling simpler, more compact and reduced-weight imaging systems, and (b) multiple-aperture techniques using single or multiple detector arrays combined with computational image construction; essentially this enables replacing the traditional approach of employing a single complex lens an array of simpler lenses to achieve functionality that is not otherwise achievable, such as extreme foveal ratios, very compact imaging systems and reduced weight.
Wavefront coding
We have demonstrated that wavefront coding can provide a SWaP-C advantage but with a penalty of degraded SNR. Two potential solutions to this problem are (a) the use of variable wavefront coding using rotational phase masks (b) agile frame-to-frame integration. Both techniques have been demonstrated in principle in the visible by Harvey's group, but have yet to be applied at thermal IR wavelengths for practical imaging systems.
Multi-aperture imaging enables imaging systems with extreme fields of view or distortions so that magnification and angular resolution (pixel-limited) is much greater (a factor of two to five) at boresight than for the periphery. Modest degrees of distortion can be achieved with conventional optics but requires significant optical complexity and factors of >2 are not possible. We will develop theory, mathematical models and algorithms for foveal imaging systems based around the following concepts that we have proposed.