针对帧差和背景减相结合的运动检测方法存在的不足，进行了以下3个方面的改进：①利用灰度拉伸变换和结合了灰度值信息的邻域相关系数计算方法，解决了背景的误判问题；②通过在帧差和背景减相结合的策略中加入运动分析，解决了缓慢运动目标的漏检问题；③采用运行期更新法更新背景模型，避免了复杂场景下背景模型的退化。实验结果表明，改进后的方法显著改善了帧差和背景减相结合的运动检测方法在背景误判、缓慢运动目标漏检以及背景模型退化等方面存在的问题。

This paper makes improvements on motion detection method which combines frame difference and background subtraction, specifically in three areas: the use of gray transform and neighborhood correlation coefficient integrated with gray value information effectively solved the problem of misdetection of moving targets; the motion evaluation step introduced in combining strategy of frame difference and background subtraction solved the problem of miss detection of slow movement targets; and the use of running average updating method to update the background model prevented the background model from degradation. Experimental results show that the improved methods have significantly solved the problems of misdetection of moving targets, miss detection of slow movement targets and background model degradation.