CAJ | 학술논문

为了提高行人防碰撞系统的预警效果，提出在车路协同环境下结合机器视觉开展行人防碰撞。车载和路侧模块分别采用机器视觉算法检测行人，采用道路构造点渠化法来快速获取局部道路地图，通过多源融合实现同质传感器条件下行人数据的融合处理，并搭建实物实验平台，基于C＋＋开发了行人防碰撞系统并进行了行人检测及预警实地实验。结果表明，系统可以检测到多个行人，并在车载端进行数据融合识别碰撞风险，并做出及时预警。
위료제고행인방팽당계통적예경효과，제출재차로협동배경하결합궤기시각개전행인방팽당。차재화로측모괴분별채용궤기시각산법검측행인，채용도로구조점거화법래쾌속획취국부도로지도，통과다원융합실현동질전감기조건하행인수거적융합처리，병탑건실물실험평태，기우C＋＋개발료행인방팽당계통병진행료행인검측급예경실지실험。결과표명，계통가이검측도다개행인，병재차재단진행수거융합식별팽당풍험，병주출급시예경。
In order to improve warning effects of pedestrian anti‐collision system ,pedestrian anti‐colli‐sion based on the Cooperative Vehicle Infrastructure System (CVIS ) and machine vision was pro‐posed .Pedestrian is detected based on the machine vision of vehicle module and road‐side module . Combining with geometrical transformation and coordinate transformation ,the coordinate of pedestri‐an relative to vehicle is gotten .T he fusion processing of pedestrian data is realized through fusion of multiple locations and multiple methods on the environment of same characters multi‐sensor .And u‐sing road structure point canalization method ,a local road map is gotten quickly .The physical test platform was established ,and pedestrian anti‐collision system based on the VC+ + were developed . Experiment result indicates that ,the system can detect multiple pedestrians and record their motion parameters ,then make timely warning aiming at dangerous pedestrian .