In-Vehicle Camera Posture Estimation for Attachment Abnormality Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting the attachment state of in-vehicle cameras are inadequate in accurately and easily identifying abnormalities, leading to erroneous or non-detection of objects during image recognition, which can impair driving assistance functions.
Innovation Solution
A method that estimates the posture of an in-vehicle camera using optical flow of feature points on a road surface, analyzing data accumulation number, standard deviation, and deviation from calibration values to determine abnormal attachment states.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If posture estimation is performed using optical flow and line segment extraction, then the system can detect camera attachment deviations, but the complexity of the detection system increases due to multiple processing steps
Solution Approach 1:
The detection process is segmented into distinct functional modules: optical flow calculation unit, line segment extraction unit, pair group extraction unit, angle estimation unit, and abnormality determination unit. Each module performs a specific task in the posture estimation pipeline, making the complex system more manageable and maintainable while preserving measurement precision.
Solution Approach 2:
The patent introduces intermediate processing steps between image capture and final posture determination. Optical flow serves as an intermediary that transforms raw pixel data into motion information, line segments serve as intermediaries that extract geometric features, and pair groups of line segments act as intermediaries that provide robust data for angle estimation. These intermediaries bridge the gap between simple image data and complex posture parameters.
2Measurement precision
If multiple parameters (data accumulation number, standard deviation, deviation amount) are monitored for abnormality detection, then the detection accuracy improves, but the processing time and computational load increase
Solution Approach 1:
The system monitors three parameters (data accumulation number, standard deviation, deviation amount) simultaneously, which is more than the minimum single parameter needed for detection. This excessive monitoring of multiple parameters ensures high detection accuracy by cross-validating abnormality conditions from different statistical perspectives, allowing the system to distinguish true abnormalities from normal variations more reliably.
Solution Approach 2:
The abnormality determination unit continuously receives feedback from the posture estimation process and compares current measurements against historical data and calibration values. The system uses feedback loops to accumulate data over time, calculate running standard deviations, and adjust detection thresholds based on observed patterns, improving detection accuracy while managing processing load through adaptive rather than static analysis.
3Reliability
If the camera attachment deviation is small, then the driving assistance function can operate with correction, but the user cannot easily identify when readjustment is necessary
Solution Approach 1:
The system provides continuous feedback to the user through abnormality notifications when detection parameters exceed thresholds. This feedback mechanism informs users when camera readjustment is needed, bridging the gap between system operation and user awareness. The feedback includes specific information about which parameter (data accumulation, standard deviation, or deviation amount) triggered the abnormality condition, enabling users to take appropriate action.
Solution Approach 2:
The notification system uses visual indicators with different colors or intensities to represent different levels of abnormality severity. This allows users to quickly grasp the attachment state status at a glance without needing to interpret numerical data, making the system easy to operate while maintaining reliable detection of both small and large deviations.
Data Source
AI summary
An information processing device includes a controller configured to execute a posture estimation processing by extracting a pair group of line segments based on an optical flow of a feature point on a road surface from a processing target region set in a camera image captured by an in-vehicle camera and by estimating a posture of the in-vehicle camera based on an angle estimation value at which the in-vehicle camera is attached, the angle estimation value being estimated from each pair data included in the pair group. The controller is configured to determine that an attachment state of the in-vehicle camera is abnormal in response to at least one of a data accumulation number of the pair data, a standard deviation of the angle estimation value, or a deviation amount from a given calibration value relating to the posture in the posture estimation processing being out of a preset range.