Weakly-Supervised Optical Flow Estimation for Time-of-Flight
This addresses motion correction for iToF cameras, an incremental improvement over existing depth correction methods.
The paper tackles motion artifacts in indirect Time-of-Flight depth images by proposing a weakly-supervised training algorithm that trains Optical Flow networks directly on reconstructed depth without ground truth flows, demonstrating it outperforms other motion compensation techniques for various sensor types.
Indirect Time-of-Flight (iToF) cameras are a widespread type of 3D sensor, which perform multiple captures to obtain depth values of the captured scene. While recent approaches to correct iToF depths achieve high performance when removing multi-path-interference and sensor noise, little research has been done to tackle motion artifacts. In this work we propose a training algorithm, which allows to supervise Optical Flow (OF) networks directly on the reconstructed depth, without the need of having ground truth flows. We demonstrate that this approach enables the training of OF networks to align raw iToF measurements and compensate motion artifacts in the iToF depth images. The approach is evaluated for both single- and multi-frequency sensors as well as multi-tap sensors, and is able to outperform other motion compensation techniques.