MEETINGS FOR 2015-2016
Awards Meeting


Wednesday 02 November 2016
14.00 hrs
City University
London, EC1V 0HB


Programme:

14.00 - 14.14 hrs WDW 2016 Award For participation at ICVS Summer School:
Temporal, Spatial And Chromatic Sensitivity Losses Associated With Opa1 Mutations
Catarina João
14.15 - 14.30 hrs Palmer award 2016 Neural Networks for Transformation to Spectral Spaces Qianqian Pan
14.30 - 15.00 hrs Colour Mentor How to commercialize your research, a personal view Prof Graham Finlayson
15.00 - 15.40 hrs Granville- Halstead Tea  
15.40 - 15.55 hrs WDW 2016 Award For participation at ICVS Summer School:
Towards a new method for characterizing reflectance discrimination capability of human observers for light sources
Panos Andrikopoulos
15.55 - 16.10 hrs Palmer award 2015 Could not be presented previously:
When is an accurate colorimetric HDR image preferable to more visually appealing images?
Keith Findlater
16.10 - 16.30 hrs General Discussion and Concluding Remarks  

Venue:
City University, Room R101, Franklin Building, 124 Goswell Road, EC1V 7DP
This is about five minutes walk south of the main campus
- see picture and map below

Time:
14.00 hrs until 17.00 hrs

Admission:
Open to all, free

Franklin Building Map

Abstracts:

Temporal, Spatial And Chromatic Sensitivity Losses Associated With Opa1 Mutations
Majander A.1,2, João C.1, Henning G.B.1, Votruba M.3, Moore A.T.1,2,4, Yu-Wai-Man P.1,2,5*, Stockman A.1*
1UCL Institute of Ophthalmology
2Moorfields Eye Hospital, London
3School of Optometry and Vision Sciences, Cardiff University Cardiff, and Cardiff Eye Unit, University Hospital Wales, Cardiff
4Ophthalmology Department, UCSF School of Medicine, San Francisco, CA
5Wellcome Trust Centre for Mitochondrial Research, Newcastle University and Newcastle Eye Centre, Royal Victoria Infirmary, Newcastle upon Tyne

Joao Purpose: Progressive retinal ganglion cell (RGC) loss is the pathological hallmark of autosomal dominant optic atrophy (DOA) caused by pathogenic OPA1 mutations. The aim of this study was to conduct an in-depth psychophysical study of the visual losses in DOA and to infer any selective vulnerability of visual pathways subserved by different RGC subtypes.
Methods: We recruited 25 patients carrying pathogenic OPA1 mutations and 15 age-matched healthy individuals. Spatial contrast-sensitivity functions (SCSFs) and chromatic contrast sensitivity were quantified, the latter using the Cambridge-Colour-Test. In 11 patients, long (L-) and short-wavelength (S-) sensitive cone temporal acuities were measured as a function of target illuminance, and L-cone temporal contrast sensitivity (TCSFs) as a function of temporal frequency.
Results: SCSFs showed impairments with the losses increasing with spatial frequency. The highest L- cone temporal acuity fell by on average 10 Hz and the TCSFs by 0.6 log10 unit. Chromatic thresholds along the protan, deutan and tritan axes were 8, 9 and 14 times higher than normal, respectively, with losses increasing with age. S-cone temporal acuity showed the most significant age-related decline. Conclusions: Losses of (i) midget, parvocellular, (ii) parasol magnocellular and (iii) bistratified koniocellular RGCs could account for the losses of (i) high spatial frequency, protan and deutan sensitivities, (ii) high temporal frequency sensitivity, and (iii) S-cone temporal and tritan sensitivities, respectively. The S-cone-related losses showed a significant deterioration with increasing patient age and could therefore serve as potential biomarkers of disease progression.



Neural Networks for Transformation to Spectral Spaces
Q. Pan*a, P. Katemakeb and S. Westlanda
aColour and Imaging Group, School of Design, University of Leeds, Leeds, United Kingdom; bColour Science Research Unit, Department of Imaging and Printing Technology, Faculty of Science, Chulalongkorn University, Bangkok, Thailand

PanThis work is concerned with mapping between the CMYK colour space and spectral space using Artificial Neural Networks (ANNs). The dimensionality of the spectral space is high (typically 31) leading to a large number of weights (or free parameters) in the network. This paper explores the hypothesis that a computational advantage can be obtained, in these cases, by treating the reflectance at each wavelength as being independent of the reflectance at any other wavelength; the implication of this hypothesis is that instead of using a single large ANN, it is possible to use, for example, 31 separate networks, each of which maps to one dimension of the 31-d spectral space. The results showed that as the number of training samples is reduced the advantage of the population of single-wavelength networks over the standard neural network approach increased.
Keywords: colour space conversion, Artificial Neural Networks (ANNs), CMYK, printing.



How to commercialize your research, a personal view
Findlayson Prof Graham Finlayson
Professor of Computing Science, School of Computing Sciences, University of East Anglia
https://www.uea.ac.uk/computing/people/profile/g-finlayson#overviewTab

Perhaps all PhD students, at some point, muse on how to make money from their research. Most PhD programs in most universities will, for example, incorporate a few talks about entrepreneurship. But, how do you go about taking an idea you have been working on in the lab and make money from it. Is it a hard or easy thing to do? (and should you want to do it at all?).
In this talk I will reflect on my own entrepreneurial journey. I graduated with a PhD in 1995 when I also won my first academic position. Since then till now I have won funding from and consulted with major international companies and spun out 4 companies. IP that I and my team have produced are in well over 100 million products. But, it hasn't been easy and there are definitely things I would - with the benefit of hindsight - have done differently. Maybe you can avoid the same pitfalls.
As part of this talk, I will reveal, in a single word, the key attribute of a successful entrepreneur.



Towards a new method for characterizing reflectance discrimination capability of human observers for light sources
Panos Andrikopoulos

Panos Colour and illumination science are relatively novel scientific fields. As a result, there is a lack of consensus in various areas with colour rendering being the most important one. Characterizing light sources in terms of their colour rendering performance is a crucial task for manufacturing and selecting light sources for many applications, such as illuminating museum, retail, residential and other environments. When illuminating artworks and especially painting, achieving colour appearance aligned with the artist's intention is the most important parameter of the illumination along with avoiding light damage. Colour rendering is consisted of three related parameters: colour fidelity, colour discrimination and appeal. Colour fidelity is assessed mainly by CIE Colour Rendering Index and appeal is considered on an experiential basis, however due to the lack of a consensus for a colour discrimination index, colour discrimination is disregarded.
LED technology has made possible the customization of light sources to maximize their performance for any of those main characteristics. This has been the case for the illumination of the Sistine Chapel where a multiband LED system using a combination has been used optimized specifically to the pigments used for the frescos and for the illumination of Mona Lisa in the Louvre Museum where the spectral characteristics of the light source where decided in situ jointly by the curators and the lighting scientists.
In this work, colour discrimination is discussed with two separate aims:
• Assessing the performance of the existing Colour Discrimination Indices.
• Assessing colour discrimination as a function of the chromaticity coordinates of the light source.
Understanding how well do existing colour discrimination indices perform, will allow decision making on which indices are appropriate to assess the performance of light sources for lighting specification in museum and other environments."



Presentation Title: When is an accurate colorimetric HDR image preferable to more visually appealing images?
Keith Findlater

Findlater High dynamic range (HDR) imaging is a process that captures a larger range of light intensities in scenes than in conventional digital photography, for which detail can be lost in highlight and shadow areas. Colour management is a process that makes an accurate transform of colour values between image devices, for example from a digital camera to a display monitor.
This research project explores techniques to capture images of scenes that have a higher subject luminance range than the dynamic range of the image sensor in a conventional digital camera. An approach is to combine standard images of different exposures of the same scene, and make a composite HDR image that records the full tonal range of the scene. The rendering of these HDR composites can produce widely differing results from lurid representations to colorimetric accurate HDR images.
How does colour fidelity change with different methods of creating HDR images? Do HDR imaging techniques produce more accurate colour reproduction than standard digital photography? What are the merits when comparing accurately colour-rendered images with subjectively enhanced images? How does HDR Image Rendering Intent affect Aesthetic Preferences?
These research questions help to ascertain the general circumstances of when an accurate colorimetric reproduction is preferable to the use of HDR imaging techniques, to generate more visually appealing images. These are investigated by comparing quantitative experimental work under controlled lighting conditions using a new colour chart, with subjective perceptual analysis of the aesthetics of the scene and image.


DATE 26 October 2016