一种应用于运动车辆检测的背景更新方法

Background Update Method Applied in Moving Vehicle Detection

基于分块思想的选择性背景更新方法存在块与块之间衔接不自然、抗干扰性弱等不足。为此,提出一种引入滤波和逐渐修正的背景更新方法。利用分块思想将图像分成区域小块,根据均值和方差确立前景区域、背景区域及其变化大小,对前景区域和背景区域分别进行更新。实验结果证明,应用该方法进行运动车辆检测,其检测正确率能提高9%左右,时间效率也稍有提高。

] The block-based selective background update method exists weaknesses that it is unnatural between the block and the adjacent block and the anti-interference is weak. Thus, the block-based selective background update method is combined with the filtering method and the gradually correction method in order to refresh the background. The current frame and the previous one are divided into small blocks. According to the mean and variance of the current background, it is established that the foreground blocks, background blocks. The background blocks and the foreground blocks are separately updated. Test results demonstrate that when the method is applied to moving vehicle detection, the accuracy is improved by about nine percent and time efficiency also improved slightly.