Markov随机游走和高斯混合模型相结合的运动目标检测算法

针对高斯混合算法对每一像素与它前后帧的像素相关联,并未考虑与相邻像素之间的关联,无法准确地捕捉到运动物体轮廓的情况,提出一种基于混合高斯模型和Markov随机游走的运动目标检测算法。利用混合高斯模型计算像素之间的颜色信息,采用Markov随机游走提取图像的边缘信息,并与提取的运动初始目标进行与计算,同时利用高斯混合模型更新背景信息。结果表明,本方法比传统的混合高斯方法具有较高的分割精度,很好的解决了混合高斯算法边缘模糊的问题,探测率也大大的提高了。

Moving target detection algorithm combined with Markov random walk and Gauss mixed model

According to the pixel of Gauss mixed algorithm is associated with its before and after the current frame, did not consider the relevance among neighboring pixels, cannot accurately capture the moving object contour, a moving target detection method based on mixed Gauss model and Markov random walk is proposed in this paper. It uses mixed Gauss model to calculate the color information between the pixels, and uses Markov random walk to extract edge information of the image to calculate with the initial moving target which is extracted. At the same time, using mixed Gauss model updates the background information. The results show that this method has higher segmentation accuracy than conventional mixed Gauss model,it is a good solution to the mixed Gauss algorithm which has edge blurring problem, the detection rate is greatly improved.