A vehicle identity determination method and system

CN122929332APending Publication Date: 2026-10-16AI SUPER EYE TECH CO LTD
0 Cites 0 Cited by

Patent Information

Application Number
CN202610791076.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-06-03
Publication Date
2026-10-16

AI Technical Summary

Technical Problem

现有方法过度依赖全局特征或车牌信息，在仅能捕捉到车顶、引擎盖等局部区域时，难以建立有效的特征匹配

Benefits of technology

[0011] Using the vehicle identity determination scheme provided above, this application embodiment constructs a vertical hierarchical priority by extracting instance masks, accurately analyzing the spatial occlusion relationship between multiple vehicles in complex scenes, and accordingly performs targeted pruning on the disturbed contour sequence to reconstruct a clean contour, effectively overcoming the morphological distortion and misjudgment problems caused by target overlap. Simultaneously, multi-dimensional heterogeneous features are extracted based on this clean contour, and an adaptive fusion model is used to dynamically adjust the weights of each feature component, enabling the system to flexibly adjust the dependence on different features according to real-time imaging quality. This significantly improves the accuracy and robustness against interference in vehicle identity determination under complex environments such as dense occlusion and poor lighting.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The application discloses a vehicle identity determination method and system. The method comprises the following steps: acquiring an image containing a plurality of vehicle targets, extracting an instance mask and a contour sequence of each vehicle target, wherein the vehicle targets comprise a target to be determined and interference targets; establishing a vertical layered priority according to the instance mask of each vehicle target; performing directional pruning on the contour sequence according to the vertical layered priority to reconstruct a pure contour of the target to be determined; extracting a multi-dimensional heterogeneous feature vector according to the pure contour and the instance mask; and outputting an identity determination result by dynamically adjusting the weight of each component in the multi-dimensional heterogeneous feature vector by using an adaptive fusion model. The scheme of the application effectively overcomes the shape distortion and misjudgment problem caused by target overlap, and can flexibly adjust the degree of dependence on different features according to real-time imaging quality, thereby significantly improving the accuracy and anti-interference robustness of vehicle identity determination.
Need to check novelty before this filing date? Find Prior Art