A multi-camera joint calibration method suitable for container first-layer loading

Abstract

This invention discloses a multi-camera joint calibration method adapted for container first-level loading, comprising the following steps: S1, setting the static calibration environment for the cameras; S2, setting the dynamic calibration environment for the cameras; S3, the cameras acquiring images of the ground; S4, preprocessing the acquired images to obtain features; S5, labeling the features and establishing a training model; S6, calculating the basic static extrinsic parameters using the training model; S7, calculating the dynamic loss compensation function using the training model; S8, dynamically adjusting the camera's extrinsic parameters using the basic static extrinsic parameters from step S6 and the dynamic loss compensation function from step S7. This invention achieves automatic calibration of multi-camera system parameters and extrinsic parameters, while ensuring high calibration accuracy, robustness, and low calibration cost.

Images