基于计算机视觉的车辆及车位检测系统设计

Design of Vehicle and Parking Cell Detection System Based on Computer Vision

针对停车困难、车位管理混乱的问题,设计了一种低成本的基于计算机视觉的车辆及车位检测系统,依靠普通监控摄像机获取停车场监控视频,采用基于Haar特征的Adaboost算法及Hough变换分别实现图像中车辆及车位的检测。实验在不同环境及不同光照条件下取得的检测正确率及召回率均超过90%。

We designed a low-cost vehicle and parking cell detection system based on computer vision for solving the problem of hard parking and parking management confusion. The common surveillance cameras were used to capture the video of parking place in this system and the vehicles and parking cells were detected by the Haar-Adaboost and Hough transformation. The experiments under different backgrounds and illumination conditions showed that both the detection accuracy and the recall rate were higher than 90%.