基于GA-BP的安全带佩戴识别方法  

Recognition method of safety belt wearing based on GA- BP

作　　者：葛如海[1] 胡满江[1] 张学荣[1] 苏清祖[1] 

机构地区：[1]江苏大学汽车与交通工程学院,江苏镇江212013

出　　处：《江苏大学学报（自然科学版）》2014年第2期125-131,共7页Journal of Jiangsu University：Natural Science Edition

基　　金：江苏省普通高校研究生科研创新计划项目(CXLX12_0628)

摘　　要：为了避免不规范佩戴安全带行为的发生,进一步提高安全带的佩戴率,提出了一种基于GA-BP的安全带佩戴识别方法.该方法在图像处理技术的基础上,提取安全带极坐标转化后的二值化图像像素值作为表征安全带佩戴状态的特征向量,并通过PCA方法对其进行降维;然后采用BP神经网络算法,建立基于BP神经网络的安全带佩戴识别模型,同时为了提高安全带佩戴识别模型的精度,引入遗传算法对其权值和阈值进行优化,建立基于GA-BP的安全带佩戴识别模型;最后通过具体实例验证.结果表明:该方法合理有效,能较好地对安全带的不同佩戴状态进行识别,具有较好的实用性和推广性.To avoid the nonstandard safety belt wearing and improve the wearing rate of safety belt, the recognition method of safety belt wearing was proposed based on GA - BP. The binarized image pixel va- lues from safety belt polar coordinates were extracted by image processing technique as characteristic vec- tor to represent the safety belt wearing state, and the dimension was reduced by PCA method. The BP neural network algorithm was used to establish recognition model of safety belt wearing. To improve the accuracy of recognition model, the genetic algorithm was used to optimize weighted value and threshold value, and the recognition model of safety belt wearing was also built based on GA - BP. The proposed models were verified by practical examples. The results show that the method is reasonable and effective, and can be used to recognize different wearing states of safety belt with good applicability.

关 键 词：安全带佩戴 状态识别 特征向量 PCA降维 BP神经网络 遗传算法 

分 类 号：U461.91[机械工程—车辆工程]