快速行人越界检测算法研究

Abstract

本文提出了一种综合运用混合高斯模型前景检测方法、Canny边缘检测、霍夫直线检测、颜色识别与重心检测方法，能较好的解决视频监控系统中行人越界检测的工程类问题。实验结果表明，此综合运用检测方法简单有效、易于实现，且检测准确率高、运行速度快。 This paper presents a comprehensive method which combines Gaussian mixture model based on the foreground detection method with Hough line detection, color recognition and centroid detection methods to detect whether the pedestrian crosses the specific boundary in the video surveillance system. Experiments show that the proposed integrated method is simple yet efficient and easy to implement, and has characteristics of high accuracy, fast running speed.