Abstract: This paper proposes an algorithm of moving objects detection in natural environment. The algorithm is based on spatial-temporal information fusion and feature recognition, including background analysis, foreground extraction, shadow removal and background updating. By analyzing the change rules of pixel parameters in HSV color model within a time period and referring to the determination formulas, it can determine whether the pixel belongs to the background, the moving objects or the shadow. The algorithm can not only eliminate the impact of natural conditions, like wind, sunlight and lightning, but also automatically update the background when the illumination changes suddenly, or moving objects stop moving, or the objects in the background turn to move. So the algorithm is suitable for moving objects detection in natural environment.

Key words: background modeling, background subtraction, shadow removal, background updating, feature extraction

摘要： 提出了一种自然环境下运动物体的监测算法，该算法基于时空信息融合与特征识别，主要包括背景分析、前景提取、去除阴影、背景更新。其思想是将图像序列均转换为HSV颜色模型，并分析比较各像素点参数在某一时间段内的变化规律，通过判定公式的判定，便可区分出各像素点在某一帧中是属于背景点、运动物体点还是阴影点。该算法针对风、阳光、闪电等自然条件可能带来的影响进行了改进，并能够在光照突变、运动物体静止后融入背景、背景物体转为运动等情况下智能更新背景，适用于自然环境下运动物体的监测。

关键词: 背景建模, 背景差分, 去除阴影, 背景更新, 特征提取