摘要
针对目前电梯门控制装置的缺陷,提出了一种基于图像识别技术的电梯门控方法,用于在电梯轿厢门区范围内实时检测运动物体.由于门区范围存在动态变化的复杂背景,图像识别易受光线干扰,该方法通过Otsu阈值分割法实现图像二值化处理,采用帧间差分法识别门区间运动物体.提出了统计的双阈值法解决门位置误差带来的问题,并提出了改进的电梯门控方法.实验结果表明:上述方法在电梯门区图像识别过程中取得了良好的效果,为改进的电梯门控方法的实现提供了可靠的保障.
For the current shortcomings of elevator door control devices,an image recognition method for the elevator door control is presented for real-time detection of moving objects in the elevator door area.As there is the complex dynamic background and it is jammed easily by light in this area,the method achieves image binarization segmentation by Otsu＇s threshold segment method,and adopts inter-frame difference method for moving object recognition.A statistical double threshold value method is proposed to eliminate the inaccuracy error of door position.At last,an improved elevator door control method is presented.Experimental results show that the above method works well in the course of image recogmtion in the elevator door area.It provides a reliable guarantee for the improved elevator door control method.
出处
《信息与控制》
CSCD
北大核心
2011年第2期243-247,共5页
Information and Control
关键词
Otsu阈值分割法
帧间差分法
统计的双阈值法
改进的电梯门控方法
Otsu＇s method
inter-frame difference method
statistical double threshold value method
improved elevator door control method
