基于混合高斯模型与三帧差分的目标检测算法

针对传统目标检测方法中光照变化、复杂背景、阴影等难点,提出了一种结合三帧差分法和混合高斯背景建模的算法,既能很好地适应场景中的光照渐变和背景扰动,又能克服普通帧差法中检测目标不准确,容易产生孔洞及双影现象的问题。同时,采用了一种简易的阴影抑制算法和形态学滤波处理,有效地去除了阴影以及噪声。实验结果表明,该算法易于实现,具有较好地实时性和鲁棒性,能精确地检测出运动目标。

An object detection algorithm based on Gaussian mixture models and three-frame difference

Aiming at the difficulties of the illumination change,complex background and shadow in the traditional target detecting methods,a novel algorithm which combines three-frame difference with mixture Gaussian background models is presented.It not only can satisfy the illumination changes and background scene disturbance,but also can overcome the problems of object detection inaccuracy which is prone to produce the voids and double-shadow phenomenon existing in common frame differencing method.Furthermore,this paper uses a simple shadow restaint algorithm and morphological filtering processing,which effectively reduces the shadows and noise.The experimental results show that the algorithm is easy to implement,has good real-time performance and robustness,and can detect the moving targets accurately.