As raised in paper,a good background model must have the following features:high precision; with the two meanings of accuracy in shape detection and reactivity to changes in time; flexibility in difference lighting conditions; and efficiency in order to provided in real-time. Many background subtraction methods have been proposed in the past decades including Running Gaussian Average , Temporal Median Filter , Mixture of Gaussians , Kernel Density Estimation,Kalman Filter , and Cooccurence of Image Variations .Among these methods, Mixture ofGaussian may be the most wildly used one. Although GMM is a robust method for background modeling, some shortcomings were discussed in some literatures and the corresponding improving methods were raised.the author pointed out two weaknesses Firstly, if the first value of a given pixel is a foreground object, there would be only one Gaussian where its weight equals unity, thus, it will take a long time until the genuine background can be considered as a background and a longer time until it will be the dominant background component.The model would not work well in busy environments because there is nearly no clean background. The author used an expected sufficient statistics update equations to begin their estimating of the Gaussian mixture model, then switch to L-recent window version when the first L samples are processed, and got a good performance. Secondly, p is too small due to the likelihood factor. This leads to too slow adaptations in the means and the covariance matrices,therefore the tracker can fail within a few seconds after initialization. They simply cut out the likelihood term from p to solve this problem. the author applied this model in different color spaces, compared their performances and concluded that YCbC is the most suitable color space for segmentation. the author incorporates the background reconstruction and foreground mergence time control into the adaptive Mixture of Gaussians, in order to construct a static background image from the video sequence and make the foreground mergence time adjustable and independent of the model's learning rate. the author built a Bayesian Framework based on the Mixture of Gaussian Models to decide a pixel belonging to foreground or background.Shadow elimination has become an active researching area in recent years. Many shadow elimination methods were proposed in literatures. the author built a GMM in YCbCrcolor space. According to the relationship of YCbC and RGBwhich described in the following equation:

GMM proposed in tends to describe a background pixel using a mixture of K Gaussian distributions. Thus, it can deal with more complex background scenes, such as waving leaves and flapping flags. At time t, the probability of observing a background pixel X, is the weighted sum of the Kdistributions:

不好意思 内容很多 跪谢了~~~~~~~~`

额~~确实好多~~ ^o^~


As raised in paper,a good background model must have the following features:high precision; with the two meanings of accuracy in shape detection and reactivity to changes in time; flexibility in difference lighting conditions; and efficiency in order to provided in real-time.
正如资料中所述，一个完善的背景模型必须具有以下几个特征：高精确度、结构把握的准确性、根据变化进行实时调整的机动性、在不同光照条件下运用的灵活性以及便于达到实时呈现效果所需的时效性。
Many background subtraction methods have been proposed in the past decades including Running Gaussian Average , Temporal Median Filter , Mixture of Gaussians , Kernel Density Estimation,Kalman Filter , and Cooccurence of Image Variations .
在过去的数十年，已有很多有关减背景技术（背景差）的方法得以应用，其中包括运用高斯均值滤波器（高斯均衡）、时间中值滤波器、混合式高斯滤波器（高斯混合）、核密度估计聚类计算法、卡尔曼滤波计算法以及共生图像转化法。
Among these methods, Mixture ofGaussian may be the most wildly used one.
然而，在这些方法中，混合式高斯滤波器运用的尤为普遍。
Although GMM is a robust method for background modeling, some shortcomings were discussed in some literatures and the corresponding improving methods were raised.
尽管高斯混合模型（GMM）对于背景规划来说是一个不错的方法，但仍然有一些文献指出了它存在的一些缺点，并相应的提出了改进方法。
the author pointed out two weaknesses Firstly, if the first value of a given pixel is a foreground object, there would be only one Gaussian where its weight equals unity, thus, it will take a long time until the genuine background can be considered as a background and a longer time until it will be the dominant background component.
作者首先指出了它的两个缺点,如果一个已知特定像素的初值是一个前景物体，那么只会有一个平衡整体的高斯点，这样的话，便需要很长时间才能使得真正的背景被识别出来，并且需要更长的时间使之变得更加明确化。
The model would not work well in busy environments because there is nearly no clean background.
由于没有清晰的背景，所以，这个模型将不能很好的在这个复杂的环境下得以呈现。
The author used an expected sufficient statistics update equations to begin their estimating of the Gaussian mixture model, then switch to L-recent window version when the first L samples are processed, and got a good performance.
作者用一系列的修正方程式统计数据开始了他们对混合高斯模型的估算，然后将初步得出的数据转换为滤镜视窗版本进行查看，效果还不错。
Secondly, p is too small due to the likelihood factor.
第二，由于某些因素影响，像素值太小。
This leads to too slow adaptations in the means and the covariance matrices,therefore the tracker can fail within a few seconds after initialization.
这就会导致距阵多个中心点匹配过慢，并且导致初始化后几秒钟内搜索追踪系统中断。
They simply cut out the likelihood term from p to solve this problem.
他们仅仅是从说明像素到解决该问题中截取了某个层面。
the author applied this model in different color spaces, compared their performances and concluded that YCbC is the most suitable color space for segmentation.
作者在不同的色彩空间运用该模型，并且将得到的结果进行对比，最后，总结出，YCBC颜色格式最适合分割色彩空间。
the author incorporates the background reconstruction and foreground mergence time control into the adaptive Mixture of Gaussians, in order to construct a static background image from the video sequence and make the foreground mergence time adjustable and independent of the model's learning rate.
作者将背景重组，并由高斯混合滤波器来控制前景合并时间，以便从视频序列中创建一个静止背景图像，同时使前景合并时间得到有效的调整并独立于模型的读取速率。
the author built a Bayesian Framework based on the Mixture of Gaussian Models to decide a pixel belonging to foreground or background.
这个作者基于高斯混合滤波器创建了贝叶斯框架（Bayesian Framework），来决定一个像素点应该属于前景，还是应该属于背景。
Shadow elimination has become an active researching area in recent years.
近年来，阴影消除理论已经成为一个活跃的研究领域。
Many shadow elimination methods were proposed in literatures. the author built a GMM in YCbCrcolor space.
很多阴影消除方法不断在文献中涌出，这个作者在YCbC颜色格式色彩空间中创建了混合高斯模型。
According to the relationship of YCbC and RGBwhich described in the following equation:
根据下面对YCBC 和RGB两种颜色格式之间的关系的描述：

GMM proposed in tends to describe a background pixel using a mixture of K Gaussian distributions. Thus, it can deal with more complex background scenes, such as waving leaves and flapping flags. At time t, the probability of observing a background pixel X, is the weighted sum of the Kdistributions:
混合高斯模型用于描述混合高斯分布区域的背景像素点，这样，它可以应用于更复杂的背景图像，如摆动的树叶和飘荡的旗帜。
在时间T点，观察背景X像素点的概率，来衡量高斯分布概况：

好好学习，天天向上~~ 仅供参考~ ：）