基于视频的停车场车位监控算法研究

Research on video-based monitoring algorithm of parking spaces

通过对各种车位图像处理的检测算法的分析和对比,选择最适合的算法在FPGA内嵌的NiosII软核中实现车位检测,该算法可以准确和快速的反映车位状态变化,保证处理的实时性。检测算法基于3种判据:车位范围内差影的方差、车位信息和背景图像的线性相关度、硬件生成的中值滤波和Sobel算子模块计算的车位边缘密度,再根据3个判据利用Fisher算法检测车位。车位检测完成后引导车位进入停车场,并通过VGA实时显示车位情况和车位引导信息。实验表明,采用SOPC技术可以加快数据处理的速度且提高系统可靠性。

By analyzing and comparing various parking space detection algorithms,the most suitable algorithm is used to achieve parking space detection in Nios II soft core embedded in FPGA.The algorithm can accurately and quickly reflect the change of state of parking spaces and can ensure real-time processing.After detection of parking spaces,real time status of parking spaces and parking guidance information can be displayed via VGA.