Stereo Camera Exposure Shift for Low-Light Obstacle Ranging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing ADAS systems face challenges in detecting obstacles quickly under low ambient illuminance due to reduced signal-to-noise ratio (SNR) and lowered ranging accuracy with SPAD sensors, which can lead to delayed detection of sudden obstacles like jumping-in or cutting-in.
Innovation Solution
An image capturing apparatus using a combination of SPAD-DAF sensors with synchronized frame and subframe timing control, allowing for adaptive image capture and processing to maintain accurate obstacle detection even in low light conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If the frame rate is increased to shorten the detection cycle for sudden obstacles, then the detection speed is improved, but the signal-to-noise ratio decreases and ranging accuracy is lowered under low ambient illuminance
Solution Approach 1:
The patent divides the image capturing process into multiple subframes within a single frame period. Each subframe captures image data for a specific time period, allowing the system to perform ranging calculations using accumulated data from multiple subframes. This segmentation enables high frame rate operation while maintaining sufficient signal-to-noise ratio through temporal accumulation of photon signals.
Solution Approach 2:
The patent performs preliminary ranging calculations using image data from previous subframes before the current frame is fully captured. By preparing ranging results in advance using accumulated data, the system can quickly output detection results when obstacles suddenly appear, effectively reducing the detection cycle without sacrificing accuracy.
2Measurement precision
If multiple pairs of cameras with different base line lengths are arranged to ensure ranging accuracy from short distance to long distance, then the ranging accuracy is improved, but the device complexity and cost increase
Solution Approach 1:
The patent transitions from spatial dimension (multiple cameras with different baseline lengths) to temporal dimension (multiple subframes within a single frame period). By accumulating image data over time across multiple subframes, the system achieves extended effective baseline length without adding physical cameras, thereby maintaining ranging accuracy across different distances while reducing device complexity.
Solution Approach 2:
The patent makes a single camera system perform multiple functions by utilizing different subframe combinations for different ranging scenarios. The same hardware can adaptively adjust the effective baseline length by selecting and combining data from different subframes, eliminating the need for multiple specialized camera pairs for different distance ranges.
3Loss of time
If the exposure time is shortened to achieve high frame rate, then the detection cycle is reduced, but the signal-to-noise ratio is lowered and obstacle detection performance deteriorates
Solution Approach 1:
The patent segments the exposure period into multiple subframes, each with shorter exposure time enabling high frame rate. By accumulating photon signals across multiple subframes during the total frame period, the system maintains sufficient signal-to-noise ratio even though individual subframe exposure times are short, thus preserving obstacle detection performance while reducing detection cycle.
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 system ensures timely detection of obstacles by optimizing image capture and processing to maintain SNR, enabling reliable ADAS performance even in low ambient illuminance scenarios.
Implementation Method 1
a single photon avalanche diode (SPAD) sensor camera using a SPAD sensor... This SPAD sensor is a sensor using a photon counter
Implementation Method 2
a stereo camera can perform a triangulation by using a parallax of two cameras to measure a distance to the obstacle
Implementation Method 3
a camera configured to perform ranging by using a sensor in which two pixels (in an ordinary configuration, one pixel) are arranged under each of microlenses arranged on an image capturing sensor and which can acquire two images with different view points
Data Source
Figure 1
Figure 2
Figure 3A~3B
AI summary
In a case where it is determined that a ranging image in which an object is not to be detected exists and it is determined that a situation where an automobile currently travels is a situation where jumping-in or cutting-in is likely to occur, an exposure start timing of one of cameras (100A, 100B) constituting a stereo camera is shifted.