Stereo Camera Calibration via Shared Model Pattern

The patent segments the calibration process into two distinct phases: (1) model pattern generation phase where a universal model is created from reference images, and (2) parameter setting phase where detection parameters are automatically determined for each camera using the pre-generated model. This segmentation allows the time-consuming model generation to be performed once, while subsequent camera calibrations benefit from the reusable model, thereby reducing overall calibration time while maintaining detection precision.

The patent performs preliminary action by generating the model pattern in advance before actual calibration begins. The model pattern, which contains essential geometric and visual characteristics of the target mark, is created once and stored for reuse. This preliminary generation eliminates the need to repeatedly create model patterns during calibration, significantly reducing calibration time while ensuring consistent detection precision across multiple cameras.

The patent applies local quality by customizing detection parameters for each camera based on its specific characteristics. While the model pattern itself is universal and reused across all cameras, the detection parameters (such as exposure time, gain, and detection thresholds) are individually optimized for each camera's distortion characteristics and performance capabilities. This ensures that each camera operates at optimal precision while benefiting from the efficiency of a shared model pattern.

The patent utilizes parameter changes by adjusting detection parameters for each camera based on the universal model pattern. The system automatically determines camera-specific parameters by comparing reference images with the model pattern, allowing each camera to have optimized detection settings that account for its unique distortion characteristics. This parameter customization maintains detection precision while preserving the efficiency gains from using a single model pattern.

The patent implements self-service by enabling the system to automatically determine and set detection parameters for each camera using the pre-generated model pattern. The calibration process requires minimal operator intervention, as the system autonomously compares reference images with the model pattern and configures appropriate parameters for each camera. This automation maintains detection accuracy while significantly improving ease of operation and reducing calibration complexity.

The patent applies universality by creating a single model pattern that serves all cameras in the system. This universal model contains the essential characteristics of the target mark and can be reused across multiple cameras with different distortion characteristics. The model pattern functions as a common reference for generating detection parameters for all cameras, reducing operational complexity while maintaining the ability to achieve accurate detection for each individual camera.