基于块匹配置信度的隧道交通背景提取算法

Background Extraction Algorithm for Tunnel Traffic Based on Confidence of Block Matching

针对隧道交通环境变化特点,采用改进的统计直方图法进行背景提取,并提出一种基于块匹配置信度的背景更新算法。实验结果表明,该算法既能满足缓慢变化场景的更新,又克服了当场景中出现慢行目标或目标长时间停留出现的拖尾和目标混入的现象,而且计算复杂度低,背景提取迅速,适合交通状况实时检测的要求。

Due to the changing characteristics of tunnel traffic environment, the improved statistical histogram method is used for background extraction, and a new algorithm based on the confidence of block matching for background updating is presented. The experimental results show that this algorithm can satisfy the slowly changing scene updating, and overcome the phenomenons of trailing when there are slow objects or objects stopping for a long time in the scence. This algorithm has lower complexity, requires less calculation and has good real-time performance for tunnel traffic detection.