Depth Image Processing for ToF Sensor Error Smoothing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Time of flight (ToF) sensors experience significant depth measurement errors due to uncertainties, leading to approximately 1% measurement errors within the measurement range, which are not adequately addressed by existing offline calibration methods, resulting in inconsistent depth value calculations.
Innovation Solution
A depth image processing method that involves obtaining consecutive depth image frames, determining trusted pixels based on preset strategies, calculating time and content similarity weights, and performing filtering processing to smooth the depth values of trusted pixels using these weights, ensuring time consistency and reducing measurement errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If offline calibration is used to correct measurement errors, then the measurement process is simplified, but the depth measurement precision remains poor with about 1% error
Solution Approach 1:
The patent applies preliminary action by performing offline calibration in advance to obtain correction parameters, which are then used during online depth measurement. This separates the complex calibration process from the real-time measurement process, simplifying the measurement operation while maintaining correction capability.
Solution Approach 2:
The patent changes parameters by introducing multiple correction parameters (k0, k1, k2, k3) that adjust the depth measurement formula dynamically. These parameters are determined through offline calibration and applied during online measurement to compensate for systematic errors, thereby improving precision without complicating the real-time measurement process.
2Device complexity
If fixed depth measurement error smoothing is applied, then processing is simple, but the depth value consistency deteriorates due to large randomness in errors
Solution Approach 1:
The patent applies dynamics by transitioning from fixed error smoothing to dynamic error correction. The correction parameters are determined through offline calibration and then dynamically applied during online measurement based on actual depth values and confidence levels, allowing the system to adapt to varying measurement conditions and improve consistency.
Solution Approach 2:
The patent implements feedback by using confidence levels as a basis for selective correction. Pixels with confidence above a threshold receive full correction, while those below receive partial or no correction, creating a feedback mechanism that adapts the smoothing intensity to the quality of each measurement point.
3Measurement precision
If multiple depth image frames are processed with time and content similarity weights, then depth measurement precision improves, but the processing complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the depth image into multiple processing stages: obtaining consecutive frames, determining trusted pixels, calculating time similarity weights, calculating content similarity weights, and performing weighted smoothing. This segmented approach breaks down the complex processing into manageable steps that can be implemented systematically.
Solution Approach 2:
The patent introduces another dimension by processing multiple depth image frames in the time dimension rather than analyzing a single frame. By incorporating temporal information from consecutive frames and calculating time similarity weights, the system enhances depth measurement precision through multi-dimensional analysis.
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
The method effectively reduces depth measurement errors by smoothing depth values based on time and content similarity weights, achieving more accurate and consistent depth calculations, particularly in areas with slow depth changes, while maintaining the dynamic nature of rapid depth changes.
Implementation Method 1
the ToF sensor determines a distance between the sensor and an object by calculating the time of flight of a pulse signal
Implementation Method 2
the ToF sensor determines a distance between the sensor and an object by calculating the time of flight of a pulse signal, and then determines a depth value of the object based on the distance
Data Source
AI summary
Provided are a depth image processing method, a depth image processing apparatus, an electronic device and a readable storage medium. The method includes: (101) obtaining consecutive n depth image frames; (102) determining a trusted pixel and determining a smoothing factor corresponding to the trusted pixel; (103) determining a time similarity weight; (104) determining a content similarity; (105) determining a content similarity weight based on the content similarity and the smoothing factor; and (106) performing filtering processing on a depth value of the trusted pixel based on all time similarity weights and all content similarity weights.