一种改进高斯模型的人体运动图像跟踪算法

摘要

Research the accurate tracking of human body based on images. This paper put forward a Gaussian mixture model of human motion tracking algorithm to complete the human tracking. A mapping was used to make the boundary of different plants in the neighborhood maximizing. After a updating of dynamic pixel, the influence of the pixel reducing caused by image rapidly changing was eliminate. The experimental results show that the improved algorithm can improve the tracking accuracy.%研究改进的人体图像序列跟踪优化问题,提高跟踪的准确性.针对当人体运动图像序列帧中,背景图像帧变化较快,帧间像素差异较大的情况下,传统的跟踪方法无法适应较大幅度的、快速的图像背景帧变化,提取的动态像素过少,导致出现跟踪前景误判,跟踪效果滞后、失准的问题.为了解决上述问题,提出一种基于高斯混合模型的人体运动跟踪算法,通过建立一种混合高斯模型模拟去除干扰,通过迭代动态像素参数更新,消除图像帧快速变化造成的像素减少带来的影响,解决跟踪滞后问题.实验结果表明,改进方法能够大幅提高在复杂背景下的人体运动准确性,取得了不错的效果.