一种改进的基于混合高斯模型的运动目标检测方法

混合高斯模型方法被广泛应用于运动目标检测，但是现有的混合高斯模型方法在应对噪声和光照突变时效果不佳，其运动目标检测的效果会受到严重影响。为了解决上述问题，提出了一种结合边缘混合高斯模型方法以及改进的基于邻域差方法的综合方法。该方法充分利用了边缘图像对于噪声和光照突变不敏感，以及邻域差方法可以去除部分噪声的特点，对图像序列进行综合处理。实验结果证明，该方法可以提升运动目标的检测率，降低误警率，可以更有效地应对噪声和光照突变的干扰，从而具有更准确的目标检测效果。

Improved moving object detection method based on Gaussian mixture mode

Gaussian mixture model (GMM) is widely used in moving objects detection in image sequences; however, the existing moving objects detection methods which are based on Gaussian mixture model are not so efficient especially when dealing with noise or illumination mutation. In order to solve the problems above, a new method is proposed. Edge images are not sensitive to noise or illumination mutation, so that the edge Gaussian mixture model (EGMM) which is constructed with edge images can be used to improve the moving objects detection. And neighborhood-based difference method is improved to restrain noise effectively. The proposed method combines the edge Gaussian mixture model with the improved neighborhood-based difference method. Experimental results prove that the proposed method can increase detection rate while reducing false alarm rate and has stronger capacity of restraining noise and dealing with illumination mutation.