(a). Imaging model for ToF cameras: a RF (f) signal modulates the amplitude of the illumination beam from the laser diode and the detector with an arbitrary phase shift. (b). Imaging model for Michelson Interferometry: the beam (ν) from the laser is the illumination in the system, and the modulation frequency is the optical frequency ν. (c). Imaging model for our proposed setup: two lasers with frequencies of ν1 and ν2 are used as illumination sources. A ‘new’ optical frequency of (ν1 − ν2) is generated after the post-processing unit, which enables micro depth resolution with macro imaging range. (d). Depth resolutions and image ranges for these three different imaging models. BS: beam splitter. R(ν): beam reflected from the mirror. S(ν): beam reflected from the object.
Project Description
Time-of-flight (ToF) sensors offer a promising method of 3D imaging due to the compact size and low complexity. However, state-of-the-art ToF sensors only have depth resolutions of centimeters due to limitations in the modulation frequencies that can be used. In this project, we propose a technique to break the limit and generate modulation frequencies as high as 1 THz using optical interferometry. Moreover, our proposed system provides great flexibility in imaging range and resolution. Our prototype demonstrates an increase in depth resolution by an order of magnitude relative to currently available commercial ToF cameras. Our prototype also demonstrates a good tolerance to the environmental vibration.
Publications
“SH-ToF: Micro Resolution Time-of-Flight Imaging with Superheterodyne Interferometry”
Fengqiang Li, Florian Willomitzer, Prasanna Rangarajan, Mohit Gupta, Andreas Velten, Oliver Cossairt
IEEE International Conference on Computational Photography (ICCP), May 2018
[pdf] [supp] [bib] [ICCP 2018 Slides]
Codes and Datasets
Coming soon
Images
Superheterodyne Interferometry
Two lasers with close wavelengths (v1 and v2) are used as illumination source simultaneously. Interference signal for each laser is detected at the detector. A processing unit is then used for post processing to generate an optical beat-note frequency (v1-v2) as the new modulation frequency in the time-of-flight system.
Comparison between our SH-ToF and regular ToF camera
This is a comparison between our proposed SH-ToF, and Michelson Interferometry and ToF cameras. As we can see, SH-ToF provides much higher depth resolution compared to regular ToF, and also larger imaging range compared to Michelson Interferometry. SH-ToF serves as a niche between these two 'time-of-flight' based imaging techniques. Please note SH-ToF is able to scan optical rough surface while Michelson Interferometry fails.
Upper-bound depth resolution with different SNR
We simulated the upper-bound depth resolution based on our physical setup. This provides the relation between depth resolution and the SNR of the signal. Different optical-beat frequency are used and compared. As we can see, higher the modulation frequency, better depth resolution.
Prototype
Two tunable lasers are used in the prototype. An APD is used as the detector. We use galvo mirrors to finish the 3D scanning.
Line scanning with different frequencies
This shows a line scanning on a planar surface. A distorted profile is observed due to the optical geometry. Different frequencies are tested in the experiment. Higher frequency provides better depth resolution but shorter image range. On the contrary, smaller frequency provides larger imaging range but worse depth resolution. One advantage of our SH-ToF, the users can choose the image range and depth resolution based on their application with the tunability of our system.
3D scanning result
This is a 3D scanning result of a plaster of David with SH-ToF.
SH-ToF vs Regular ToF
This provides the comparison of scanning results from SH-ToF and regular ToF. As we can see, more details such as the eye ball and eyelids can be visualized, which is totally missed in regular ToF scanning results.
Acknowledgments
This project was partially supported by DARPA REVEAL program, NSF CAREER IIS-1453192, ONR N000141612995, and ONR N00014-15-1-2735.
