2021
DOI: 10.1109/tci.2021.3126533
|View full text |Cite
|
Sign up to set email alerts
|

iToF2dToF: A Robust and Flexible Representation for Data-Driven Time-of-Flight Imaging

Abstract: Indirect Time-of-Flight (iToF) cameras are a promising depth sensing technology. However, they are prone to errors caused by multi-path interference (MPI) and low signal-to-noise ratio (SNR). Traditional methods, after denoising, mitigate MPI by estimating a transient image that encodes depths. Recently, data-driven methods that jointly denoise and mitigate MPI have become state-of-the-art without using the intermediate transient representation. In this paper, we propose to revisit the transient representation… Show more

Help me understand this report
View preprint versions

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

0
15
0

Year Published

2021
2021
2024
2024

Publication Types

Select...
2
2
2

Relationship

1
5

Authors

Journals

citations
Cited by 23 publications
(15 citation statements)
references
References 54 publications
0
15
0
Order By: Relevance
“…To conclude, we will now see how our model fares in the presence of extremely high amounts of shot noise. To this aim, we trained our approach on the iToF2dToF dataset [23], which, as described in Section V, has a very high amount of zero-mean errors. To put things into perspective, the single frequency reconstruction at 100 MHz of the test set from the measurements with shot noise leads to a MAE of 7.24 cm, while the same computation done on images with only MPI, gives an error of 1.45 cm, meaning that MPI accounts only for 20% of the total reconstruction error.…”
Section: B Results On Mpi Correctionmentioning
confidence: 99%
See 4 more Smart Citations
“…To conclude, we will now see how our model fares in the presence of extremely high amounts of shot noise. To this aim, we trained our approach on the iToF2dToF dataset [23], which, as described in Section V, has a very high amount of zero-mean errors. To put things into perspective, the single frequency reconstruction at 100 MHz of the test set from the measurements with shot noise leads to a MAE of 7.24 cm, while the same computation done on images with only MPI, gives an error of 1.45 cm, meaning that MPI accounts only for 20% of the total reconstruction error.…”
Section: B Results On Mpi Correctionmentioning
confidence: 99%
“…from [23], we used two input frequencies, 20 and 100 MHz, masked the edges using a Canny edge detector during testing and did not consider the highest 1% of errors for the final computation. Our approach shows some remarkable denoising capabilities even in this scenario as it can be seen in Figure 6.…”
Section: B Results On Mpi Correctionmentioning
confidence: 99%
See 3 more Smart Citations