Abstract

ViBe algorithm is a powerful moving object detection algorithm. It has many advantages, such as simple method, easy implementation and high computational efficiency, but there are also many shortcomings, such as ghost problem, susceptibility to noise and illumination changes and inadaptability to dynamic scenes. Aiming at the above shortcomings, an adaptive ViBe algorithm based on dispersion coefficients and spatial consistency factor is proposed. Firstly, the mode of multi-frame images is used to replace single image to realize initialization, which reduces the interference of ghost to background model; Secondly, dispersion coefficient is used to establish adaptive dynamic threshold to improve the adaptability of the algorithm to dynamic background; Finally, the spatial consistency factor with spatial information is used to establish adaptive update factor, which reduces the error rate and enhances the robustness of the algorithm. The experimental results show that our improved ViBe algorithm can effectively eliminate ghosts, better adapt to noise, illumination and dynamic background, have more complete detection results and higher detection accuracy than the traditional and others’ improved ViBe algorithms and Gaussian mixture model.

中文翻译：

---

基于色散系数和空间一致性因子的自适应Vibe算法

摘要

ViBe算法是一种功能强大的运动物体检测算法。它具有许多优点，例如方法简单，易于实现和计算效率高，但是也存在许多缺点，例如重影问题，对噪声和光照变化的敏感性以及对动态场景的不适应性。针对上述缺点，提出了一种基于色散系数和空间一致性因子的自适应ViBe算法。首先，采用多帧图像模式代替单幅图像实现初始化，减少了幻影对背景模型的干扰。其次，利用色散系数建立自适应动态阈值，以提高算法对动态背景的适应性。最后，利用具有空间信息的空间一致性因子建立自适应更新因子，降低了错误率，提高了算法的鲁棒性。实验结果表明，与传统的ViBe算法和其他高斯混合模型相比，改进后的ViBe算法能够有效消除重影，更好地适应噪声，光照和动态背景，检测结果更加完整，检测精度更高。