|
There are many applications of 3D displays where the viewer needs to look at transparent shapes, one example is in medical imaging where doctors need to make a diagnosis using a 3D scan of a patient, another is airport security where security staff must look at x-ray images of luggage to check for weapons.
Our exhibit at the Royal Society is designed to help us understand whether there is a difference in everyone's ability to spot weapons in a transparent image when it is viewed in either 2D or stereoscopic 3D.
To do this we are using the specially created Spot the Weapon game* where a sequence of computer graphics images are shown to viewers and they must make a decision whether each image contains or does not contain a weapon, for example one of the above images does and one does not.
What do we expect ?
This is the first step for us investigating this research topic so we don't yet know what to expect - which is why we need your help - but:
-we think there may be a difference in the time it takes to determine if a weapon is present in a 2D image compared to a stereoscopic 3D image.
-we think there may be a difference in the number of missed weapons and false alarms when we compare results between 2D and stereoscopic 3D images.
During the week we will anonymously collect results from everyone who agrees to take part and an analysis of all the results will be uploaded here.
If you took part thank you for your contribution to this important research, if you have any questions or thoughts please email the address below.
Analysing the data from your results
These are preliminary results - more to follow as we analyse the data in more detail.
Monday 4th: 16 people took part (on press day)
While this is a small number of people it is enough to analyse the data and one significant result is:
In scenes where a weapon is present : on average people were 7.5% less accurate in detecting the weapon when viewing images in stereoscopic 3D compared to viewing in 2D.
Tuesday 5th: 81 people took part (full day open to everyone)
The data today looks a little noisy - however when we analysed the data one significant result is:
In scenes where a weapon is present: on average people were 2.5% more accurate in detecting the weapon when viewing the images in stereoscopic 3D compared to viewing in 2D.
The results are proving challenging to analyse, for example people are helping each other with the answers and this is giving some great variation in results. More to follow as we figure this one out.
Technical design of the exhibit
The software developed for this exhibit is written in Python, this is a modern rapid prototyping computer language, allowing us to quickly develop and test different approaches to presenting the images and collecting your data. When the exhibition is over the software will be made freely available on these web pages. It was developed using the Enthought Python software distribution.
The 3D display we are using is a Hyundai 24" 3D display this is HD quality and uses circular polarisation to separate the left and right images. This allows us to use exactly the same 3D glasses made by RealD as you use for watching 3D movies like Avatar in the cinema.
*Throughly tested by Guy, Susannah and Naomi.
| 
