3D Scene Inference from Transient Histograms

Jungerman, Sacha; Ingle, Atul; Li, Yin; Gupta, Mohit

doi:10.1007/978-3-031-20071-7_24

Cited by 6 publications

(7 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A challenge, addressable by pixelated ToF sensors and minimized by small pixel FoVs, is the differentiation between subsurface scattering delays and the spreading of the transient return due to object tilt. Transient returns captured by the AMS TMF8801 were shown to allow inference of the object tilt [ 25 ]. Another model simplification instance is when materials are considered highly scattered and diffusive.…”

Section: Discussionmentioning

confidence: 99%

Plastic Classification Using Optical Parameter Features Measured with the TMF8801 Direct Time-of-Flight Depth Sensor

Becker

Koerner

2023

Sensors

View full text Add to dashboard Cite

We demonstrate a methodology for non-contact classification of five different plastic types using an inexpensive direct time-of-flight (ToF) sensor, the AMS TMF8801, designed for consumer electronics. The direct ToF sensor measures the time for a brief pulse of light to return from the material with the intensity change and spatial and temporal spread of the returned light conveying information on the optical properties of the material. We use measured ToF histogram data of all five plastics, captured at a range of sensor to material distances, to train a classifier that achieves 96% accuracy on a test dataset. To extend the generality and provide insight into the classification process, we fit the ToF histogram data to a physics-based model that differentiates between surface scattering and subsurface scattering. Three optical parameters of the ratio of direct to subsurface intensity, the object distance, and the time constant of the subsurface exponential decay are used as features for a classifier that achieves 88% accuracy. Additional measurements at a fixed distance of 22.5 cm showed perfect classification and revealed that Poisson noise is not the most significant source of variation when measurements are taken over a range of object distances. In total, this work proposes optical parameters for material classification that are robust over object distance and measurable by miniature direct time-of-flight sensors designed for installation in smartphones.

show abstract

Section: Discussionmentioning

confidence: 99%

Plastic Classification Using Optical Parameter Features Measured with the TMF8801 Direct Time-of-Flight Depth Sensor

Becker

Koerner

2023

Sensors

View full text Add to dashboard Cite

show abstract

“…A competing authentication method using 3D facial recognition has emerged in the mobile device market: 26 a full 3D direct time-of-flight (dTOF) imager. 27 The required detector elements employing a multichannel 3D dTOF, comprised of a vertical-cavity surface-emitting laser (VCSEL) array 28,29 and accompanying single photon avalanche diode (SPAD) sensor, [30][31][32] can be manufactured small enough for integration into a display to enable such an advanced feature.…”

Section: Time-of-flight Sensingmentioning

confidence: 99%