基于图像的车祸事故人工智能识别系统的设计  

Artificial Intelligence Recognition System Design of the Accident Based on Image

作　　者：倪燃[1] 

机构地区：[1]山东交通学院信息科学与电气工程学院,济南250357

出　　处：《科技通报》2013年第8期166-168,共3页Bulletin of Science and Technology

基　　金：山东交通学院科研基金(Z201211)

摘　　要：当前的交通车祸监测系统进行识别时,无法处理交通环境的复杂干扰因素,出现车辆混合和误判问题,不能准确获取车祸区域特征,降低了交通车祸事故识别的准确性。提出一种基于车辆图像灰度内方差的道路交通车祸识别方系统的设计方法,给出了系统设计思路,系统获取道路交通监测范围的图像,经过核心芯片运算获取道路图像的灰度类内方差,获取道路交通车祸图像特征,分析车辆图像特征的变化系数,最终完成对道路车祸事故的准确识别。实验结果说明,该系统能够及时准确识别出相应道路的车祸事故,具有较高的精确度,识别误差率较低,具有较强的应用价值。Monitoring system by the current traffic accident identification, unable to process complex traffic environment interference factors, hybrid vehicle and misjudgment problems, can not accurately obtain accident area characteristics, re- duce the traffic accident accident identification accuracy. Based on vehicle image gray-level variance within a road traf- fic accident identification system design method, gives the system design thought, system for road traffic monitoring range of image, through the core chip operation for road image gray-level variance within the class, obtaining image character- istics of road traffic accident, analysis the change coefficient of vehicle image characteristics, finally complete the accu- rate identification of road accident accident. Experimental results indicate that the system can accurately identify the cor- responding road crash accident in time, with high accuracy, the recognition error rate is low, has strong application value.

关 键 词：道路交通事故 车祸识别 灰度内方差 图像特征 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]