CAJ | 학술논문

　　针对目前车辆安全预警系统因价格高、系统复杂和虚警频繁等难以普及推广的现状，提出一种能够快速智能地识别出障碍物是实警还是虚警的新模型，以排除虚警对系统的干扰。该模型利用双CCD立体摄像头作为传感器，无论车辆直线行驶还是转弯行驶，根据车辆当前行驶状况，都能够提前预测车辆即将行驶的运动轨迹线，从而使检测到的不在轨迹线内的虚警障碍物得以排除，提高了系统的鲁棒性、安全性、智能性。通过 Matlab 对图像中障碍物进行虚警、实警的模拟仿真识别和判断，达到了预期要求。
　　침대목전차량안전예경계통인개격고、계통복잡화허경빈번등난이보급추엄적현상，제출일충능구쾌속지능지식별출장애물시실경환시허경적신모형，이배제허경대계통적간우。해모형이용쌍CCD입체섭상두작위전감기，무론차량직선행사환시전만행사，근거차량당전행사상황，도능구제전예측차량즉장행사적운동궤적선，종이사검측도적불재궤적선내적허경장애물득이배제，제고료계통적로봉성、안전성、지능성。통과 Matlab 대도상중장애물진행허경、실경적모의방진식별화판단，체도료예기요구。
Aiming at the current situation where it's difficult to promote and widely use the existing vehicle safety warning system due to its high price, complicated system, and frequent false-alarm, etc., the paper puts forward a new model which can rapidly and intelligently identify whether the barrier to be real -alarm or false-alarm, to eliminate the inter-ference of false-alarm to the system.With double CCD stereo camera as the sensor , the model can predict the motion trajec-tories in advance that vehicle is about to run according to the vehicles'current driving situations, for either straight driving or turning driving, so that the detected false-alarm barriers beyond the motion trajectories can be eliminated , which can improve the system robustness, safety and intelligence.Through simulated identification and judgment of real-alarm or false-alarm for barriers in the image of Matlab , the expected requirement is reached.