Camera Calibration Adjustment Using Object Movement Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing camera calibration methods for autonomous driving systems are prone to errors due to variations in camera position and angle, leading to inaccuracies in distance and speed calculations for target objects.
Innovation Solution
An electronic device that utilizes deep learning-based 2D bounding box information and keypoint information to determine movement information of target objects and adjust camera calibration information accordingly, thereby compensating for slight camera warping and ensuring reliable vehicle control data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If camera calibration information is obtained through physical calibration at the EOL vehicle production stage, then the calibration information can be obtained for general mass-produced vehicles, but the camera position or angle may vary due to internal and external factors causing errors in distance and speed calculations
Solution Approach 1:
The system performs preliminary physical calibration at the EOL vehicle production stage to obtain initial calibration information, but then continuously performs automatic calibration during vehicle operation to compensate for subsequent position or angle variations of the camera, thus maintaining measurement precision while accounting for reliability concerns about calibration stability
Solution Approach 2:
The system uses feedback from image recognition results and movement information of target objects to automatically adjust and update calibration information during vehicle operation, creating a closed-loop system that continuously corrects calibration errors caused by camera position or angle variations, thereby improving both measurement precision and reliability
2Measurement precision
If physical camera calibration is performed to ensure accurate distance calculations, then the calibration information can be obtained, but the process requires significant time and effort
Solution Approach 1:
The system performs a single preliminary physical calibration at the EOL vehicle production stage to obtain initial calibration information, avoiding the need for repeated manual calibration processes, thus reducing calibration time while maintaining measurement precision through subsequent automatic adjustments
Solution Approach 2:
The system enables self-service automatic calibration during vehicle operation by using image recognition results and movement information to automatically adjust calibration information without requiring manual intervention, thereby eliminating the need for repeated time-consuming manual calibration processes while maintaining accurate distance calculations
Data Source
AI summary
An electronic includes a communication device that receives object information of a target captured by a camera and recognized based on a deep learning model. The electronic device also includes a memory storing calibration information of the camera and physical information of the target. The electronic device additionally includes a processor configured to determine movement information of the target based on the object information. The processor is also configured to determine a first physical quantity to the target based on the movement information and the calibration information. The processor is further configured to determine a second physical quantity to the target based on the movement information and the physical information. The processor is additionally configured to compare the first physical quantity with the second physical quantity to determine whether to adjust the calibration information.