基于三帧间的移动障碍物背景目标提取算法

Algorithm of Moving Obstacle Background Object Extraction Based on Three-frame Difference Method

为了减少交通事故的发生及降低因其而带来的危害,针对障碍物检测过程中出现的各种问题。受原有的两帧间差分法的启发,提出了一种基于三帧间快速匹配检测障碍物的方法,该算法利用移动障碍物这一特性进行障碍物的检测。其基本思想是利用相邻三帧图像两两差分,再将两个差分结果再次进行差分操作从而最终确定运动目标在图像中的位置,实现了对移动障碍物轮廓的检测。

In order to reduce the occurrence of traffic accident and the losses caused by it,aimed at the problems that occurred in the process of obstacle detection.Inspired by the original two-frame difference method,a new fast and robust obstacle detection method which is based on three-frame difference is proposed in this article.The algorithm uses the mobile feature to detect the obstacle and the basic idea is to use the neighboring three frames to make difference with each other,then reuse the results of the two differential images to make difference once again in order to confirm the target position in the image and finally successfully detect the adumbration of the moving obstacle.