基于隔帧差分和背景减去法的运动车辆检测算法

Moving vehicles detection algorithm based on discontinuous frame-difference and background subtraction

为了解决单一检测算法在复杂场景中不能完整检测出运动车辆的问题,提出一种复杂场景中运动车辆的检测算法。该算法利用当前较为新颖的稀疏去噪方法对检测的视频序列进行去噪,然后采用一种改进统计平均法初始化背景,并结合隔帧差分和背景减法对运动车辆进行检测,最后应用一种新的双权值更新策略对背景进行更新。实验表明,该算法建立的背景可靠,抗干扰性强,改善了车辆的检测效果,具有很好的实用价值。

To solve the single detection algorithm can＇t complete detect moving vehicles in complex scene, a complex scene moving vehicle detection algorithm is presented. Video image denoising through sparse denoising method has been well acknowledged as an novel approach of sequence image of video de- noising in recent years. Detection algorithm using a kind of improved statistical - averaged method to ini- tialize background, combine discontinuous frame - difference with background subtraction, and detect moving vehicles. Finally, a new double -weights update strategy was used to update the background. Ex- perimental results show that the establishment of background has reliability and robust. Detection algo- rithm improves the detection effect of the vehicle and it has the very good application value.