ToF Pulse Histogram Correction for Long Tail Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Time of Flight (ToF) devices struggle with accurately identifying the presence of objects due to the presence of undesirable long tail distributions in the returned pulse histograms, which can lead to false positives and reduced frame rates, especially at close distances.
Innovation Solution
A method is proposed to precompute a global pulse shape into a reference part and a correction part, aligning the reference part with the current histogram to determine the long tail part, and then subtract or compensate for the long tail distribution using the precomputed correction part.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional identification algorithms are used to detect objects in ToF histograms, then the detection process is simple, but false positives increase due to long tail distributions
Solution Approach 1:
The patent segments the histogram analysis into two distinct parts: the useful pulse region and the long tail region. By applying different processing strategies to each segment (thresholding for useful pulses, modeling for long tail), the system achieves accurate object detection while managing computational complexity effectively.
Solution Approach 2:
The patent introduces an intermediary mathematical model (Gaussian distribution for useful pulse, exponential distribution for long tail) that acts as a mediator between the raw histogram data and the final detection decision. This model-based approach enables reliable differentiation between true object signals and long tail artifacts.
2Reliability
If long tail bins are ignored to reduce false positives, then detection reliability improves, but secondary pulses in the tail are occulted
Solution Approach 1:
The patent converts the harmful long tail distribution into a beneficial signal by modeling it with an exponential distribution. Instead of treating long tail bins as noise to be discarded, the system characterizes their expected distribution and uses this knowledge to identify genuine secondary pulses that deviate from the modeled long tail pattern.
Solution Approach 2:
The system uses feedback by continuously comparing the observed histogram bins against the modeled long tail distribution. This comparison provides feedback that enables the system to distinguish between expected long tail artifacts and genuine secondary pulses, allowing accurate detection even in the tail region.
3Measurement precision
If complex algorithms are used to remove long tail distributions, then detection accuracy improves, but frame rate decreases
Solution Approach 1:
The patent changes the approach from complex iterative long tail removal algorithms to a parameter-based modeling approach. By pre-defining mathematical models with specific parameters (Gaussian for useful pulse, exponential for long tail) and fitting them to the histogram data, the system achieves high detection precision with computationally efficient operations that maintain high frame rates.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach effectively removes or compensates for the long tail distribution, simplifies pulse detection, maintains high frame rates at close distances, and ensures consistent frame rates across varying distances with low computational complexity.
Implementation Method 1
emitting means ME configured to emit optical radiations RE
Implementation Method 2
A photon may generate a carrier in the SPAD through the photoelectric effect
Implementation Method 3
measuring the time taken by the signal to travel to the object and to travel back to the device
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
Method for detecting a presence of at least one object (OBJ) within a field of view of a time-of-flight sensor (SENS), comprising: - obtaining (410) a current histogram (HST) generated by the time-of-flight sensor, the histogram comprising a distribution of bins associating a number of detected photons to a given acquisition time, - providing (420) a precomputed nominal global pulse shape of the distribution, including a reference part (421) and a correction part (422); - performing (430) a fitting of the position of at least the reference part (421) to the current histogram (HST); - determining (440) the position (FT_423) of a long tail part (LT) of the current histogram using the position (423) of the correction part (422) relatively to the reference part (421) inside the precomputed nominal global pulse shape; - calculating an estimate of a long tail distribution in the current histogram from the correction part (422) and the determined position (FT_423); and - performing a correction, (600) in the current histogram, of the calculated estimate of the long tail distribution.