说明：  自适应核密度估计运动检测方法 提出一种自适应的核密度(kernel density estimation, KDE)估计运动检测算法. 算法首先提出一种自适应前景、背景阈值的双阈值选择方法, 用于像素分类. 该方法用双阈值能克服用单阈值分类存在的不足, 阈值的选择能自适应进行, 且能适应不同的场景. 在此基础上, 本文提出了基于概率的背景更新模型, 按照像素的概率来更新背景, 并利用帧间差分背景模型和KDE分类结果, 来解决背景更新中的死锁问题, 同时检测背景的突然变化. 实验证明了所提出方法的适应性和可靠性.
(Adaptive kernel density estimation method of motion detection of an adaptive kernel density (kernel density estimation, KDE) estimated motion detection algorithm. Algorithm first proposed an adaptive prospects, the background of the dual-threshold threshold selection method for pixel classification. The method used to overcome the dual-threshold single classification threshold shortcomings, the choice of threshold can be adaptive, and able to adapt to different scenes. On this basis, this paper, based on the probability of the background update model, in accordance with the probability of pixels to update the background, and take advantage of interframe differential background model and the classification results of KDE, to resolve the deadlock in the background update, while detection of abrupt changes in the background. Experiments show the proposed method of adaptability and reliability.)