摘要
为克服现有的安全带报警系统存在的不足,提出了一种安全带佩戴视觉检测系统并通过MATLAB实现.该系统首先把采集到的彩色图像转化为灰度图像并进行预处理,然后设计了合适的感兴趣窗口(W0I),用该WOI对灰度图像进行截取,以削减图像数据运算量,同时有效排除了其他区域的干扰,增强了安全带特征.选取合适的全局阈值进行图像分割,计算得到的二值图像中的亮点比率,与设定的亮点比率阈值进行比较,从而判别出安全带是否佩戴规范.再利用MATLAB R2012a软件开发了安全带佩戴视觉检测系统软件,实现了对安全带佩戴规范与否的检测.最后,试验选取不同乘员身穿不同衣物,在不同光照环境下进行图像采集与检测试验,总体正确识别率达98.3%.试验结果表明,该检测系统快速有效,具有较强的鲁棒性和实时性.
In order to solve some existing problems in safety belt warning system, a vision detection system is put forward and simulated in MATLAB. Firstly, a collected color image is converted into a grayscale one, and an appropriate Window of Interest （WOI） is designed for cropping the grayscale image, with the purpose of reducing the image data computation, eliminating the interference from other regions and enhancing the characteristics of safety belt. Secondly, an appropriate global threshold for image segmentation is select, the bright spot ratio in the binary image is computed, and the normalization of wearing the safety belt is judged comparing with the setting threshold of bright spot ratio. Then, a vision detection system for safety belt wearing status is developed based on MATLAB R2012a platform. Finally, an experiment of image acquisition and detection is carried on with different occupants, wearing different clothes, under different lighting conditions. The correct recognition rate reaches 98.3%. The experimental results indicate that the detection system works quickly and efficiently, and has a strong robustness.
出处
《车辆与动力技术》
2013年第4期17-21,25,共6页
Vehicle & Power Technology
关键词
安全带
佩戴状态
视觉检测
感兴趣窗口
机器视觉
Safety beh
Wearing status
Vision detection
Window of interest
Machine vision
