A vehicle re-identification optimization method and device based on extreme lighting conditions

By constructing a simulated dataset and an illumination recovery network, and combining the recovery and recognition association network to optimize the vehicle re-identification model, the problem of insufficient accuracy in vehicle re-identification under extreme lighting conditions was solved, and high-precision vehicle identification under extreme lighting conditions was achieved.

CN118411680BActive Publication Date: 2026-05-29WUHAN UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
WUHAN UNIV
Filing Date
2024-04-03
Publication Date
2026-05-29

AI Technical Summary

Technical Problem

Existing technologies are affected by the accuracy of vehicle re-identification under extreme lighting conditions, especially in low light and high exposure conditions where vehicle feature extraction is inaccurate, leading to inaccurate identification results.

Method used

A simulated dataset was constructed, including image data under various lighting conditions. Illumination restoration was performed using the Extreme Illumination Estimation and Recovery (EIER) framework. The vehicle re-identification model was optimized by combining the Reconstruction and Recognition Association Network (RIAN). The reconstruction network and re-identification process were trained using the illumination-restored image set. Association parameterization was established to improve recognition accuracy.

Benefits of technology

It effectively solves the accuracy problem of vehicle re-identification under extreme lighting conditions, improves the accuracy of vehicle image retrieval under extreme lighting conditions, and enhances the accuracy and consistency of identification.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118411680B_ABST
    Figure CN118411680B_ABST
Patent Text Reader

Abstract

The application relates to the field of computer vision and re-identification technology, in particular to a vehicle re-identification optimization method and device based on extreme light conditions. The method comprises the following steps: constructing a simulation data set according to a VehicleID data set and a VERI-Wild data set, wherein the simulation data set comprises image data under multiple light conditions; classifying and performing light recovery processing on the images under extreme light conditions in the simulation data set to generate a light recovery image set; training a re-recovery network for re-identification; cascading the re-recovery network and a re-identification process; and training the cascaded process by taking the light recovery image set as input to complete vehicle re-identification optimization. In the embodiment of the application, the simulation data set comprising multiple light conditions is constructed, and the extreme light pictures in the simulation data set are classified and recovered to train a vehicle re-identification model, thereby solving the problem that the extreme light has a great influence on the re-identification technology in the related art.
Need to check novelty before this filing date? Find Prior Art