东华大学学报（自然科学版）

ViBe算法是一种基于静态背景下的运动目标检测算法，针对其“鬼影”问题和运动目标静止时会被更新为背景的问题提出了改进ViBe算法，即对原ViBe算法的背景模型初始化、动态阈值、前景分割和背景模型更新等4个部分进行了改进。采用均值法获取的背景图像初始化背景模型，可消除“鬼影”;利用计数法控制前景分割动态阈值，使前景图像更加准确；使用帧差法思想改进前景分割，使前景图像更加完整；通过引入阈值保证背景模型更新的稳定性。根据试验结果可知，改进ViBe算法对正常移动车辆、较小运动目标和存在静止情况的运动目标都有较好的检测能力，解决了“鬼影”问题和运动目标静止时会被更新为背景的问题，同时相较于原ViBe算法和其他常用运动目标检测算法，改进ViBe算法在保证准确性的基础上提高了检测的完整性。

The ViBe algorithm is a moving target detection algorithm based on static background. Aiming at its "ghost" problem and the problem that moving targets will be updated as background when they are stationary, an improved ViBe algorithm is proposed, which improves the original ViBe algorithm in four parts: background model initialization, dynamic threshold, foreground segmentation and background model update. The background image obtained by the mean method is used to initialize the background model and eliminate the "ghost". The dynamic threshold of foreground segmentation is controlled by counting method to make the foreground image more accurate. Then the frame difference method is used to improve the foreground segmentation and make the foreground image more complete. The threshold is introduced to ensure the stability of background model updating. According to the test results, the improved ViBe algorithm has good detection ability for normal moving vehicles, small moving targets and moving targets with static conditions, and solves the problem of "ghost" and the problem that moving targets will be updated as background when they are stationary. At the same time, compared with the original ViBe algorithm and other common moving target detection algorithms, the algorithm improves the integrity of detection on the basis of ensuring the accuracy.