U.S. flag An official website of the United States government

U.S. Food and Drug Administration

Point Spread Function Method for Spatial Resolution of Augmented Reality Head Mounted Displays

Catalog of Regulatory Science Tools to Help Assess New Medical Devices 

This regulatory science tool (RST) is a laboratory method for measuring the spatial resolution of optical see through augmented reality head mounted displays (AR HMDs) using point spread function analysis.

Technical Description

This regulatory science tool (RTS) is a method for measuring the point spread function (PSF) and calculating the modulation transfer function (MTF) of an augmented reality head mounted display. Spatial resolution is an important image quality parameter for many applications of medical extended reality (MXR). The spatial resolution is frequently measured using a grille pattern or a sinusoidal pattern at different spatial periodicities to measure the contrast modulation function (CTF) or the MTF, respectively. For head mounted displays (HMDs), grille patterns are more commonly used due to the HMD pixels. However, this approach requires repeated measurements with test patterns at different spatial frequencies and the grille pattern provides the CTF opposed to the MTF. Compared to grille methods, the line spread function (LSF) and PSF methods allow for determining the spatial resolution and MTF from a single measurement. LSF gives the spatial resolution and MTF in one dimension, whereas PSF gives the two-dimensional (2D) spatial resolution and 2D MTF. However, due to the subpixel structure of many virtual reality displays, the PSF method is only valid for optical see-through AR HMD.

The tool specifies the following information: 

  1. Test pattern: Render a 3 x 3 array of dots that are one-pixel wide or as small as possible through the rendering pipeline centered with respect to the optical axis of each eye piece of the HMD. This can be accomplished by positioning the test pattern in the center of the field of view (FOV) with a large enough virtual target distance that the vergence angle is approximately 0o (≥ 100 m). The inter-dot spacing should be at least a factor 10 larger than the measured dot full width at half maximum (FWHM) to ensure each dot can be measured independently. In addition, the angular width of the 3 x 3 array should be ≤ 2o with an inter-dot spacing between 0.5o-1o, corresponding to the fovea of the human visual system. This pattern allows for the 2D spatial resolution or angular resolution to be measured at a single location in the FOV from a single measurement. The details are described in a peer-reviewed publication [1] and Appendix A. The RST Toolkit for Evaluation of Head Mounted Display Image Quality [2] can be used to render the dot pattern on an HMD with a compatible web browser.
  2. Experimental setup: This method uses a high-resolution monochromatic array light measuring device (LMD) with photopic response mounted on a 5-axis goniometer or robot system to align the center of the entrance pupil of the HMD. Prior work demonstrated that eye-rotation methods, opposed to pupil rotation methods, are necessary to capture the impact of eye gaze due to the pupil shift as the eye rotates [1]. Therefore, the rotation axis should be placed approximately 10 mm behind the entrance pupil of the HMD for the eye-rotation method. Technical requirements on the LMD used are specified in IEC 63145-20-10 standard [3] and Sec. 19.2 of the Information Display Measurements Standard (IDMS) [4]. It is recommended to have at least 20:1 for the LMD to HMD pixels ratio to minimize the impact of the LMD modulation transfer function and ensure adequate sampling of the HMD pixels. Finally, the LMD should have near diffraction limited performance to minimize the impact of the LMD wavefront errors on the measurements.
  3. Image acquisition and processing procedures: Images of the rendered dots should be acquired using the LMD. The captured images can be used to determine the spatial resolution and MTF of the HMD using the procedure described in detail in [1] and Appendix A.  The nine dots in the test pattern should be analyzed independently and used to calculate the uncertainty in the measurements. The measurement may be repeated with the dots rendered at different angular locations in the FOV to determine the spatial resolution across the FOV. The measurements can also be repeated for red, green, and blue dots to measure the chromatic dependence of the spatial resolution.

Detailed step-by-step instructions on implementing this tool are described in the User Manual.

Intended Purpose 

This method is intended to characterize the spatial or angular resolution and MTF of an HMD. The method is applicable to stereoscopic optical see-through AR HMDs. The method can be implemented by medical device developers and testing houses for evaluating MXR devices.

Testing

The method has been extensively tested on multiple HMDs: Microsoft HoloLens2, Epson Moverio BT-300, and the Magic Leap 2. The tests involve repeating the measurements at different locations in the FOV of the HMD, repeating for different eye pieces of each HMD, and repeating on a conventional monitor placed at the virtual focal plane. The conventional monitor was also used to characterize the performance of the LMD.

Limitations

The PSF method requires a high-resolution array LMD, such as a camera, with near diffraction-limited performance and a 5-axis goniometer for alignment. The minimum feature size that can be rendered may depend on the rendering engine and anti-aliasing methods implemented in the rendering pipeline. This method is only intended for optical see-through AR HMDs and not intended for VR HMDs or video see-through HMDs where spatial resolution measurements are impacted by the subpixels of the HMDs.

Supporting Documentation

The method with detailed instructions and validation are described in the following publication:

[1] Johnson, M., Zhao, C., Varshney, A., & Beams, R. (2024). Quantifying the optical and rendering pipeline contributions to spatial resolution in augmented reality displays. Journal of the Society for Information Display, 32, 555-567. https://doi.org/10.1002/jsid.1297

[2] RST Toolkit for Evaluation of Head Mounted Display Image Quality, https://cdrh-rst.fda.gov/toolkit-evaluation-head-mounted-display-image-quality

Reference standard documents:

[3] IEC 63145-20-10:2019 Eyewear display - Part 20-10: Fundamental measurement methods - Optical properties.

[4] Information Display Measurements Standard, SID, 2023.

[5] Instructions on Implementing the point spread function measurements.

Contact

Tool Reference