基于灰度积分投影的安全带佩戴识别方法  

A Method of Recognizing Seat-belt Wear Based on Gray Scale Integral Projection

作　　者：葛如海[1] 胡满江[1] 符凯[1] 

机构地区：[1]江苏大学汽车与交通工程学院,镇江212013

出　　处：《汽车工程》2012年第9期787-790,共4页Automotive Engineering

基　　金：吉林大学汽车动态模拟国家重点实验室开放基金(20091106);江苏省普通高校研究生科研创新计划项目(CXLX12_0628)资助

摘　　要：为进一步提高安全带佩戴率,针对目前安全带佩戴提示系统存在的不足,提出了一种基于灰度积分投影的安全带佩带识别方法。该方法在对采集到的乘员图像进行预处理得到灰度分配较为均匀图像的基础上,采用全局阈值分割法对图像进行阈值分割得到二值化图,再对二值化图像分别进行垂直和水平灰度积分投影,结合安全带几何特征的先验知识,得到安全带特征点的位置坐标,从而达到安全带佩戴识别的目的。实验结果表明,该方法能有效识别乘员安全带佩戴状况,具有较高的适用性和准确性。To further increase the seat belt wearing rate, and in view of the inadequacy of existing prompt system for seat belt wear, a new recognition method of seat belt wear is proposed based on gray scale integral projec- tion. With the method, by preprocessing the collected occupant pictures, an image with a relatively uniform gray scale distribution is obtained, which is converted into binary images by global threshold segmentation. Then both vertical and horizontal gray scale integral projections are conducted on binary images, combined with a priori knowl- edge on the geometric features of seat-belt, to get the position coordinates of the feature points of seat-belt. Finally, the recognition on seat belt wear is realized. The results of tests show that the method proposed can effectively recog- nize the seat-belt wear state of occupants with relatively high applicability and accuracy.

关 键 词：安全带识别 特征提取 阈值分割 积分投影 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] U492.8[自动化与计算机技术—计算机科学与技术]