【摘要】 在基于视频图像处理的乘客计数系统(APC)中,背景与前景目标的分割是运动目标检测的关键。针对公交车APC系统的特征,在背景差法的基础上,结合相邻帧差以及平均灰度差,提出了一种自适应背景更新算法,以此为基础,实现了视频图像序列中的运动目标检测。通过对公交车实验采集的视频图像进行处理,证明算法能够克服光线的变化及干扰物体的影响,有效地实现目标分割。更多还原

【Abstract】 In vision-based APC system,segmentation of background and objects is the key in passenger detection.According to the character of bus APC system,this paper using both the delayed subtraction and average gray information,presents an adaptive background updating method based on background subtraction,and implements moving objects detection of video sequences successfully.By processing stereo sequences acquired from buses,the results of experiments show that this method can extract moving objects against illumination level changes and extraneous motion efficiently.更多还原