Stochastic Exposure Coding for Handling Multi-ToF-Camera Interference

Jongho Lee, Mohit GuptaICCV 2019 · OpenReview

Abstract

As continuous-wave time-of-flight (C-ToF) cameras become popular in 3D imaging applications, they need to contend with the problem of multi-camera interference (MCI). In a multi-camera environment, a ToF camera may receive light from the sources of other cameras, resulting in large depth errors. In this paper, we propose stochastic exposure coding (SEC), a novel approach for mitigating MCI. SEC involves dividing a camera's integration time into multiple slots, and switching the camera off and on stochastically during each slot. This approach has two benefits. First, by appropriately choosing the on probability for each slot, the camera can effectively filter out both the AC and DC components of interfering signals, thereby mitigating depth errors while also maintaining high signal-to-noise ratio. This enables high accuracy depth recovery with low power consumption. Second, this approach can be implemented without modifying the C-ToF camera's coding functions, and thus, can be used with a wide range of cameras with minimal changes. We demonstrate the performance benefits of SEC with theoretical analysis, simulations and real experiments, across a wide range of imaging scenarios.

Introduction and Problem Statement

Continuous-wave Time-of-Flight (C-ToF) cameras, common in mobile devices and robotics, measure depth by emitting light and analyzing the phase shift of its reflection. However, when multiple cameras operate in proximity—a scenario known as Multi-Camera Interference (MCI)—they receive light from each other's sources. This results in two distinct types of interference: AC interference, which causes phase shift errors, and DC interference, where cumulative light acts as intense ambient glare, creating significant photon noise.

Comparison of interference effects on depth signals.

Traditional solutions like orthogonal coding address AC interference but fail to mitigate the DC component. In dense environments, this extra DC light can plummet the signal-to-noise ratio (SNR), making depth recovery impossible. While Time-Division Multiple Access (TDMA) could theoretically solve this, it requires nanosecond-level synchronization across all devices, which is impractical for consumer hardware.

Stochastic Exposure Coding (SEC)

To address these challenges, we propose Stochastic Exposure Coding (SEC), a method that achieves the benefits of TDMA without requiring synchronization. SEC operates by dividing the total exposure time into numerous small slots (at the microsecond or millisecond scale). In each slot, a camera independently decides to turn "ON" with a specific probability ($p_{ON}$).

The proposed multi-layer coding architecture combining micro-scale exposure coding with nano-scale depth coding.

This approach utilizes a layered coding architecture. The Exposure Coding Layer operates at micro-time scales to handle DC interference, while the Depth Coding Layer remains at the nano-time scale to preserve standard depth measurement functions. By detecting "clashes"—slots where multiple cameras are active simultaneously—the system can discard corrupted data. Alternatively, a "Combined" (CMB) approach integrates SEC with traditional orthogonal coding, allowing for high-accuracy depth recovery even in the presence of residual interference.

Experimental Validation and Results

The performance of SEC was evaluated using simulated 3D models, a hardware prototype with four Texas Instruments C-ToF cameras, and real-world testing on complex scenes.

Key findings include:

  • Superior Accuracy: In simulations with 10 interfering cameras, SEC and the CMB approach maintained high depth accuracy (RMSE ~6mm), whereas traditional industry methods like PN-sequences saw error rates explode (RMSE ~80mm).
  • Energy Efficiency: SEC can achieve better depth results than conventional methods while using only 30% of the energy, as the camera is only active during selected slots.
  • Robustness: The system effectively filters interference even when the number of cameras is unknown, making it highly practical for uncontrolled environments.

Conclusion and Future Work

Stochastic Exposure Coding offers a robust, software-defined solution for multi-camera interference that is compatible with existing C-ToF hardware. By embracing randomness, it bypasses the need for centralized synchronization while effectively mitigating both AC and DC interference. Future research will focus on optimizing SEC for dynamic scenes to prevent motion artifacts and exploring the thermal limits of high-peak-power pulsing required for the "ON" slots.

Summary updated

Abstract

As continuous-wave time-of-flight (C-ToF) cameras become popular in 3D imaging applications, they need to contend with the problem of multi-camera interference (MCI). In a multi-camera environment, a ToF camera may receive light from the sources of other cameras, resulting in large depth errors. In this paper, we propose stochastic exposure coding (SEC), a novel approach for mitigating. SEC involves dividing a camera's integration time into multiple slots, and switching the camera off and on stochastically during each slot. This approach has two benefits. First, by appropriately choosing the on probability for each slot, the camera can effectively filter out both the AC and DC components of interfering signals, thereby mitigating depth errors while also maintaining high signal-to-noise ratio. This enables high accuracy depth recovery with low power consumption. Second, this approach can be implemented without modifying the C-ToF camera's coding functions, and thus, can be used with a wide range of cameras with minimal changes. We demonstrate the performance benefits of SEC with theoretical analysis, simulations and real experiments, across a wide range of imaging scenarios.
BibTeX
@inproceedings{lee2019stochastic,
  author = {Jongho Lee and Mohit Gupta},
  title = {{Stochastic Exposure Coding for Handling Multi-ToF-Camera Interference}},
  booktitle = {IEEE/CVF International Conference on Computer Vision},
  year = {2019},
  url = {https://openreview.net/forum?id=7OibrSxifDm},
}
Figure