Purchase this article with an account.
Chris McCarthy, Deepa Prabhu, Clare MacMahon, Marten De Man, Lisa Wise; Effect of camera position on localisation with simulated prosthetic vision. Invest. Ophthalmol. Vis. Sci. 2020;61(7):924.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Most current designs of visual prostheses use an external camera to capture information from the scene. Positioning of the external camera is an important issue since it is likely to affect the egocentric spatial frame of reference. The present simulated prosthetic vision (SPV) study investigates the direct effects of camera position on egocentric localisation by way of analysing the direction and magnitude of pointing errors and associated head movements immediately after exposure and prior to adaptation.
Fifteen sighted participants (10 right-eye viewing (REV), 5 left-eye viewing (LEV)) performed blocks of a pointing task for each of four camera positions: (i) centre mounted (CM), (ii) head mounted (HM), (iii) left mounted (LM), and (iv) right mounted (RM). Participants pointed at white circular stimuli that were randomly presented on a black background while viewing through a simulation of a 44-channel retinal implant on a head-mounted display. Actual and participant-identified stimulus locations were recorded on a touchscreen. Head and finger movements were captured using a Qualisys Motion Capture System. This study complied with guidelines of National Statement on Ethical Conduct in Human Research.
As can be seen from Table 1, in both LEV and REV participants, absolute errors were minimum for CM followed by HM position. Errors were maximum when the camera was mounted contralateral to the viewing eye. CM errors had lower standard deviation and variance than HM errors. Direction of pointing was found to be strongly correlated with direction of head orientation with high positive correlation for HM, CM (.69 ≤ r ≤ .89) and moderate to high correlation for RM and LM (.49 ≤ r ≤ .89) in both groups. Relative to other camera mount positions, the percentage of participants with significant (r ≥ .66, p= .05) correlation was higher for HM (77.7% REV, 100% LEV).
Our results showed lower errors for centrally mounted camera positions (CM and HM) versus laterally mounted with higher accuracy and precision for the CM position. High correlation between direction of head orientation and pointing with CM and HM positions suggests that centrally mounted positions facilitate intuitive head orientation cues that are important for reliable egocentric direction estimation.
This is a 2020 ARVO Annual Meeting abstract.
This PDF is available to Subscribers Only