应用改进混合高斯模型的运动目标检测

在运动目标检测过程中，背景建模对目标提取至关重要，而混合高斯模型是目前背景建模中较流行的方法之一。针对混合高斯模型中存在的不足做了两点改进：（1）混合高斯模型是对各点孤立建模，对于拥有较高的分辨率的图像运算量较大，引入分块建模思想，可以明显提高目标检测的速率而且考虑到像素点之间的空域信息；（2）混合高斯模型对运动目标停留在场景中某一位置停留过长时，会出现将前景转化成背景，以致于产生目标在场景中消失的现象，根据目标在场景中运动与静止的情况，决定是整帧更新还是只更新背景区域。通过实验可以得出，该算法在不影响识别的情况下可以显著地提高运动目标的检测速率，而且可以减少部分噪声，另外也能有效地克服目标转化为背景的情况，从而保持了运动目标出现的连续性。

Moving object detection based on improved Gaussian Mixture Models

In the course of the moving-object detection,the background models are crucially important for the object extraction,but the Mixture Gaussian Model（GMM） is one of popular methods in the background models.For the deficiency of the GMM,two improvements are proposed in this paper.Firstly,the blocking model is introduced,which can significantly improve the detection rate with the airspace information among the pixels considered, while the GMM bases on the pixels and it needs much larger operations for the high-resolution image.Secondly,the foreground will be turned into background so as to it disappears if the moving object stay on one position of the scene for a long time.The solution is to update the entire frame or just update the background according to the state of moving object.The results show that the method can not only improve the detection rate of the moving object without affecting identification and reduce some noise,but also effectively solve the problem that the object turn into the background.Thus it can maintain the continuity of moving object.