University of Birmingham > Talks@bham > Computer Science Lunch Time Talk Series > Random Matrices, Compressive Sensing, and the Light Field

Random Matrices, Compressive Sensing, and the Light Field

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Christine Zarges.

Room changed

I will introduce some of the imaging methods of current interest in Birmingham, and how computation plays a crucial role in both understanding the images themselves, and also in creating them in the first place. I will focus on three particular imaging types and their associated computations:

i) Mass Spectrometry Imaging and Random Basis Approximations Mass Spectrometry is a widely used chemical analysis technique, and modern instrumentation allows spatially-resolved spectra to be recorded, allowing the distribution of chemicals within a sample to be analysed. The data is high-dimensional and standard multivariate data analysis methods fail. We have shown how a carefully constructed random projection can be used to reduce the dimensionality of the data in a principled way, and how multivariate analysis can then be performed on the compressed dataset. Further, it is straightforward to project the results of any analysis back into the original (physical) space. This has enabled the analysis of far larger datasets than has previously been possible.

ii) Diffuse Imaging and Compressive Sensing In diffuse optics, we attempt to reconstruct the interior structure and properties of a strongly scattering object from optical measurements made from its exterior. In certain situations, such as luminescently tagged cancers, the source of light is interior, and compact. I will introduce the principles of compressive sending and show how they can be used to enhance the localisation of the source, and hence the cancer. [NOTE: this was the subject of my recent departmental seminar and there will be some repetition].

iii) Light Field Images and their reconstruction. Light field imaging is a method by which directional information can be captured by a simple camera, via a very simple modification of its optics followed by complex computation. I will show some examples of what can be done through different computations on the light field, and how we are extending this to look at some important biomedical problems.

This talk is part of the Computer Science Lunch Time Talk Series series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

Talks@bham, University of Birmingham. Contact Us | Help and Documentation | Privacy and Publicity.
talks@bham is based on talks.cam from the University of Cambridge.