Computational Imaging Group

X-ray of a chest

What is the Computational Imaging Group?

The computational imaging group conducts research and provides postgraduate training in a variety of practical imaging applications. Much of the research is concerned with image reconstruction or image recovery: computing images of objects, scenes, internal structures, etc., from a variety of data. Although the strength of the group is in the theoretical and computational aspects of imaging, the focus is on practical applications, including the design of instrumentation (particularly acoustic and optical), algorithm development, data collection and computational implementation. The application areas are primarily in acoustic, astronomical, biophysical and biomedical imaging, and remote sensing. The group is led by five academic staff, together with approximately 20 postgraduate students and a number of postdoctoral fellows and research technicians. The group works with a number of national organisations and overseas research groups.

Research Projects

Research projects in the Computational Imaging Group fall into six areas which are outlined, together with the assoicated academic staff, below. For information on specific projects, contacts the academic staff member concerned.

Synthetic aperture Sonar (SAS) is a technique for high resolution imaging of the seafloor and is the sonar analog of synthetic aperture radar. SAS attains its high along-track resolution by using a "synthetic aperture", achieved by coherent processing of reflected acoustic signals collected along the path of a towed transmitter/receiver. The performance of current SAS systems is limited by distortions due to platform motion and aperture undersampling. The objective of our research program is to develop methods to produce high quality acoustic images of the seafloor using synthetic aperture techniques. This involves development of sophisticated computer algorithms to compensate for platform motion, modelling ocean acoustic phenomena, design and construction of advanced electronic instrumentation, and underwater engineering. Experiments are conducted mainly in Lyttelton Harbour but also in Auckland and Sydney Harbours.

More information can be found on the Acoustics Research Group website.


Sample publications:

  • K.A. Johnson, M.P. Hayes and P.T. Gough, A method for estimating the subwavelength sway of a sonar towfish, IEEE J. Oceanic Eng., 20, 258-267 (1995).
  • P.T. Gough and D.W. Hawkins, Imaging algorithms for a strip-map synthetic aperture sonar: Minimizing the effects of aperture errors and aperture undersampling, IEEE J. Oceanic Eng., 22, 27-39 (1997).
  • H.J. Callow, M.P. Hayes and P.T. Gough, Wavenumber domain reconstruction of SAR/SAS imagery using single transmitter and multiple-receiver geometry, Electron. Lett., 38, 336-338 (2002).

In many non-invasive medical imaging techniques (e.g., magnetic resonance imaging, x-ray computed tomography, optical imaging, etc.), indirect measurements are made and sophisticated computational algorithms are used to form an image from these data. We are developing advanced image reconstruction algorithms to improve image quality, feature detection and classification, and computational speed. We use transform techniques, shape representation, statistical modeling, optimisation, and propagation models. Applications include fast algorithms for MRI, detecting and correcting for motion in MRI, reconstruction of x-ray CT images, and reconstruction methods in optical diffusion imaging. Our research in biomedical signal processing focuses on modelling and detection of events in physiological signals, and inverse solutions for biomedical fields. Applications include detection of drowsiness and location of electrical sources from EEGs.


Sample publications:

  • Maclaren, J.R., Bones, P.J., Millane, R.P. and Watts, R. MRI with TRELLIS: A novel approach to motion correction. Magnetic Resonance Imaging, 26 (4), 2008: 474-483.
  • Wu, B., Millane, R.P., Watts, R. and Bones, P.J. Exploiting image sparsity in parallel magnetic resonance imaging (pMRI). Image Reconstruction from Incomplete Data V, Proceedings SPIE, 7076, 707603:1-11, 2008.
  • Butler, A., Bones, P. and Hurrell, M. Prototype system for enhancement of frontal chest radiographs using eigenimage processing. J. Medical Imaging and Radiation Oncology, 52 (3), 2008: 244-253.
  • Van Hese, P., Vanrumste, B., Hellez, H., Carroll, G.J., Vonck, K., Jones, R.D., Bones, P.J., D'Asseler, Y. and Lemahieu, I. Detection of focal epileptiform events in the EEG by spatio-temporal dipole clustering. Clinical Neurophysiology, 119, 2008: 1756-1770.
  • Peiris, M. T. R., Jones, R. D., Davidson, P. R., Carroll, G. J., & Bones, P. J. Frequent lapses of responsiveness during an extended visuomotor tracking task in non-sleep-deprived subjects. Journal of Sleep Research, 15, 2006: 291-300.

Methods such as x-ray diffraction and electron and optical microscopy are used to image molecules and molecular assemblies in biological systems on µm to nm scales. Because the scattering cross section of these objects is so small, these techniques depend on the objects being arranged in regularly ordered arrays, so that the diffracted signals are coherently amplified. However, particularly for molecular assemblies with large aspect ratios, the ordering is sometimes imperfect. Our research is concerned with analysis and modeling of disordered arrays and their diffraction properties, analysis of images and diffraction patterns, and applications to polymer fibres and biomolecular arrays (e.g., microfibrils in muscle and the cone mosaic in the retina).


Sample publications:

  • W.J. Stroud and R.P. Millane, Cylindrically averaged diffraction by distorted lattices, Proc. Roy. Soc. London Ser. A, 452, 151-173 (1996).
  • R.P. Millane and A. Goyal, Analysis of the disordered myosin filament lattice in muscle, Fiber Diffr. Rev., 9, 6-11 (2000).
  • B. Bodvarsson, S. Klim, S. Mortensen, M. Morkebjerg, J. Chen, J.R. Maclaren, C.H. Yoon, P.K. Luther, J.M. Squire, A. Bainbridge-Smith, P.J. Bones and R.P. Millane, Determination of myosin filament positions and orientations in electron micrographs of muscle cross-sections, In "Image Reconstruction from Incomplete Data III," P.J. Bones, M.A. Fiddy and R.P. Millane (Eds.), Proc. SPIE, Vol. 5562, 97-108 (2004).

Aircraft and spacecraft based remote sensing provides a cost-effective means of wide area imaging of the Earth’s surface. Also, hand-held laser scanning gives three-dimensional surface profile information on geological structures with mm accuracy. The data obtained have applications in environmental studies, land use monitoring, flood control, etc. Our research program is concerned with development of methods for analysis and interpretation of remotely sensed imagery. Techniques used include probabilistic modeling (Bayesian estimation and Markov random fields), optimisation, statistical classification, and non-linear filtering. Applications are to airborne and spaceborne synthetic aperture radar for classifying types and ages of forests, and for detecting overall seasonal trends, to airborne laser altimetry for quantitating vegetation cover in the vicinity of braided rivers, and to hand-held laser scanning for analysis of pebble size distribution and modelling of gravel river beds. Another novel project involves analysis of sailplane flight data to investigate the structure of atmospheric waves in the lee of large mountain ranges.


Sample publications:

  • M. Hagedorn, P.J. Bones, Q.X. Wu and D. Pairman, Segmentation of synthetic aperture radar images. Proc. Image Vision Comput. NZ 99, 73-78 (1999).
  • T. Bretschneider, P.J. Bones and S.J. McNeill, Resolution enhancement using multispectral remotely sensed imagery. Proc. Image Vision Comput. NZ 99, 109-114 (1999).
  • P.J. Bones, T. Bretschneider, C.J. Forne, R.P. Millane and S.J. McNeill, Tomographic blur identification using image edges, Proc. SPIE, 4123, 133-141 (2000).
  • R.P. Millane, R.G. Brown, E. Enevoldson and J.E. Murray. Estimating mountain wave windspeeds from sailplane flight data. In "Image Reconstruction from Incomplete Data III," P.J. Bones, M.A. Fiddy and R.P. Millane (Eds.), Proc. SPIE, Vol. 5562, 218-229, 2004.
  • M. Qi, A. Haider, R.P. Millane and G.M. Smart. Analysis of river bed stones from digital elevation models. In "Proc. Image and Vision Computing New Zealand 2004," D. Pairman, H. North and S. McNeill (Eds.), Landcare Research, NZ, 369-373, 2004.

At the low-end of the processing scale microcontrollers are interesting examples of what could almost be called a System on a Chip, ie. a single chip which contains a complete computer system. The top-end FPGAs, for example the Xilinix Virtex II Pro, are an exciting example of a high-end processing device. The Xilinx Virtex II Pro has on-board GB/s serial drivers, CPU cores, and memory sitting in a "sea" of FPGA logic. However the configuration or programming, typically with VHDL or Verilog, of these devices is not simple, and VHDL is no longer the most suitable language. We are developing alternatives building off the current state-of-the-art in software engineering, with an emphasis on applications to image processing. This includes investigating and developing tools such as: an Aspected Oriented Language and compiler, refactoring tools, identifying hardware/software patterns and software critics, modelling and analysis using petrinets, and partitioning sequential and parallel code.


Images of astronomical objects with ground-based telescopes are blurred by refractive index variations that evolve with time (Figure 1 left). Adaptive optics is an opto-mechatronic system that can overcome the effects of the earth’s atmosphere and reduce the amount of blurring (Figure 1 right). This system (Figure 2) consists of a sensor, which measures the instantaneous wavefront distortion due to the atmosphere, an adaptive mirror that can be locally deformed by independent actuators, and a control law to calculate the optimal actuator commands from the wavefront sensor measurements and atmospheric and noise statistics.

Currently the next generation of Extremely Large Telescopes are being designed, including the European Southern Observatory’s (ESO) European Extremely Large Telescope (EELT) (Figure 3), a 39m diameter primary mirror telescope, which will be the world’s largest when finished. We are collaborating with ESO on the design of the adaptive optics systems for the EELT, including solving various signal and image processing and control problems using numerical simulations.


Sample publications:

  • R. Muradore, L. Pettazzi, R. Clare, and E. Fedrigo, "An application of adaptive techniques to vibration rejection in adaptive optics systems," Control Engineering Practice. 32, 87-95 (2014).
  • R. M. Clare, M. Le Louarn, and C. Béchet, "Laser guide star wavefront sensing for ground-layer adaptive optics on extremely large telescopes," Applied Optics 50, 473-483 (2011).
  • R. M. Clare, M. Le Louarn, and C. Béchet, "Optimal noise-weighted reconstruction with elongated Shack–Hartmann wavefront sensor images for laser tomography adaptive optics," Applied Optics 49, G27-G36 (2010).

For advice

Rick Millane

ICTS Rm 202
Internal Phone: 94536

Find research groups

Search our Research database

News, events and seminars