Method and system for vehicle micro-collision capture and trajectory reconstruction in an anti-occlusion environment

CN121920245BActive Publication Date: 2026-05-29XIAMEN UNIV OF TECH
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
XIAMEN UNIV OF TECH
Filing Date
2026-03-25
Publication Date
2026-05-29

AI Technical Summary

Technical Problem

Existing technologies struggle to detect minor vehicle collisions and reconstruct accident trajectories containing real dynamic characteristics under conditions of visual obstruction and weak collision features, leading to difficulties in determining liability.

Method used

By constructing a vehicle semantic spatiotemporal graph with dynamic occlusion perception, extracting frequency domain micro-motion features based on phase Euler amplification, and using a collision event neural inference and conditional denoising diffusion model driven by physical constraints, the vehicle's motion trajectory within the occlusion area is reconstructed.

Benefits of technology

It effectively detects minor collisions and accurately reconstructs accident trajectories, improving the accuracy of collision identification and providing strong evidence for determining liability in traffic accidents.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121920245B_ABST
    Figure CN121920245B_ABST
Patent Text Reader

Abstract

The application discloses a kind of vehicle micro-collision capture and trajectory reconstruction method and system under anti-shielding environment, comprising: constructing dynamic shielding perception vehicle semantic space-time graph, locking the visible rigid connection feature site of shielding risk target;Based on phase amplification, the feature site is analyzed in frequency domain micro-motion, and vibration fingerprint feature vector is extracted in combination with vehicle type-modal frequency fingerprint library;By introducing impulse-momentum theorem as the neural network of physical constraint, the vibration fingerprint feature is inferred and regressed collision impulse of collision event;Using the denoising diffusion model with collision impulse as the condition, the vehicle trajectory in the shielding area that meets the collision dynamics is repaired and generated.The application analyzes the micro-vibration of unshielded parts to infer collision, and integrates physical principles into neural networks, solving the problems of difficult micro-collision detection and trajectory distortion under shielding, and generating trajectories that retain collision dynamics characteristics, providing strong evidence for accident liability determination.
Need to check novelty before this filing date? Find Prior Art