基于深度学习的车辆全景影像拼接与融合算法研究

打开文本图片集

中图分类号：U463.675 文献标识码：A 文章编号：1003-8639（2026）02-0082-03

Research on Vehicle Panoramic Image Mosaic and Fusion Algorithm Based on Deep Learning

WuXiaoyu,SunJiahui (Guangzhou Automobile Group Co., Ltd., Guangzhou 51oooo, China)

【Abstract】 With the increasingly complex urban trafic environment,the integration of a panoramic image system incarshasbecomethe mainstreamsolution to solvethecollsionrisk.However,thereare alsochallengesin theapplication of this technology,such as deformation fault and color jump intheoverlappng area of viewangles.Therefore,this paper discusses the panoramic image generation technology driven by deep learning,deeply analyzes the spatial mapping relationshipof multi-camera images,anddesignsand developsamodel basedonan embedded platform.Thereal vehicle verificationshows thattheapplicationofthis model intypical parkingscenarioscan efectivelyavoidtherisk of vehicle bodyscratches,while improving theadaptability to extremeenvironmentsand hardware compatibility,significantly enhancing driving safety and technical popularity.

【Key Words】panoramic image mosaic；deep learning;performance verification;intelligent driving and asistance

0 引言

近几年来，随着人民收入水平的提高，城市车辆愈发密集，泊车事故发生频率居高不下。（剩余5045字）