Software to colour-correct mobile phone images to enable a sample’s colour change to be deduced in a number of lighting conditions, using a range of mobile phone cameras -- 2

We wish to design and build software that allows for the consistent colour correction of samples photographed by a variety of mobile phone cameras, specifically for the purpose of quantitatively measuring colorimetric test reactions (e.g. urine tests for things like glucose, in strip form) in the home. The scientific measurement of colour, using an uncontrolled camera and lighting system presents many possible sources of error (lighting levels, lighting colour, camera sensitivity, camera linearity, differential colour sensitivity, differences between cameras on what the dyes for the Bayer layer are, etc). We want to bypass the issue of every camera phone being different in terms of hardware, internal algorithms, etc by ensuring a series of standards are contained within the image field to allow for the generation of a processed, colour calibrated JPG using the data in the image itself. In this way the JPG contains all the calibration information we need (i.e. we don't actually need to know anything about the phone at all) as well as (potentially) the encoded colour / concentration dependence curve which allows for the quantification of the colorimetric reaction. Project Steps: 1. Calibration object production: Design “coloured array” of spots that will act as control and allow correction of colour variation. Calibration object to also aid with localization of test strip area. We will supply pictures and/or sizes of the read-out test result card so appropriate coloured dots can be produced 2. R&D: Investigate distribution of colours and variation between pictures taken of the same test strip under different lighting conditions with different models of mobile phone. We will supply pictures of strips taken under different conditions (on different coloured backgrounds) e.g. natural light, sunlight, soft artificial light, LED artificial light. Flash or no-flash. VSS to ideally measure actual test strip colour change (e.g. RBG levels) using lab equipment if possible. Calibration object provided in (1) to be used as the background. 3. Hypothesis proof: Prove hypothesis that we can colour correct our commercial test strips using the above method – the results from this exercise should show we get the same result from each setting (e.g. lighting condition and phone model) 4. Prototype production: Make simple "investor-ready" prototype proving front-end on either PC/Web/Mobile phone (phone preferred) that shows reading for at least one test strip on at least one model of phone 5. Estimate accuracy / sensitivity of first prototype (steps will be taken to increase accuracy in part 2 of project) DEVLIERABLES - First-protoype (for at least one mobile phone model (must be one that a company director has!) or web/PC and one of the available test stips) · All underlying code (IP must be owned by us) · Document detailing methods / how we proved our hypothesis (ppt or word), accuracy, sensitivity, ideas around next steps e.g. ways to improve accuracy and how to build upon this for more models of phone, biomarkers etc