Assessing visual attention in virtual reality: Automatic one-point calibration for eye-tracking

To assess visual attention in head-mounted displays (HMDs), a re-calibration of the eye-tracker is necessary each time the HMD is moved or taken off and put on again. This occurs several times per therapy session in some use cases of virtual reality based rehabilitation. The process is time-consuming and likely a demanding task for patients. Hence, a solution is needed which reduces the workload and does not require explicit user input for calibration while wearing a HMD. We propose a method for a one-point calibration, which is very fast, appropriately accurate, and can be performed automatically. A user study with healthy subjects showed significant improvements in eye-tracking accuracy using our method after reattaching the HMD