A method and apparatus for constructing an image velocity field dataset based on digital twins

By generating RGB three-channel images in 3D modeling using digital twin technology, the accuracy and adaptability issues of multi-degree-of-freedom target velocity field annotation in existing technologies are solved, enabling automated annotation of pixel-level velocity fields and reducing costs.

CN120032046BActive Publication Date: 2026-06-30UNIT 32002 OF THE CHINESE PEOPLES LIBERATION ARMY
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
UNIT 32002 OF THE CHINESE PEOPLES LIBERATION ARMY
Filing Date
2024-12-31
Publication Date
2026-06-30

AI Technical Summary

Technical Problem

Existing technologies struggle to accurately label the pixel-level velocity field of each instance in images of multi-degree-of-freedom, flexible, and clustered targets. Furthermore, differences in sensors lead to poor system adaptability and high labeling costs.

Method used

Using a digital twin-based approach, a scene is built through 3D modeling and simulation. An RGB three-channel image is generated using a rendering engine. Combined with the motion state settings of the instance, the position vector, linear velocity field, and angular velocity field are automatically calculated to achieve pixel-level velocity field annotation.

Benefits of technology

It improves the accuracy and adaptability of velocity field annotation, reduces annotation costs, achieves full-process automation, and avoids the time and cost of manual annotation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120032046B_ABST
    Figure CN120032046B_ABST
Patent Text Reader

Abstract

This invention proposes a method and apparatus for constructing an image velocity field dataset based on digital twins. It includes: step three, constructing a motion scene of an instance; setting instance motion information, determining the instance's position vector, linear velocity field, and angular velocity field; step four, generating an instance mask, obtaining an instance color mask map and corresponding position vector map, linear velocity vector map, and angular velocity vector map; step five, rendering the image, using a rendering engine to synthesize the image based on the object's position vector map, linear velocity field map, angular velocity field map, and instance color mask map, obtaining the instance's visible light image; and step six, setting the number of samples in the dataset, using this number as the loop count for executing steps one through four; each loop generates one instance image sample based on the instance's visible light image. This invention improves the accuracy of position vector and velocity field annotation, enhances the dataset's adaptability to different image acquisition devices, and reduces the cost and difficulty of velocity field annotation.
Need to check novelty before this filing date? Find Prior Art