基于视频图像的交通枢纽综合体人体检测

针对交通枢纽综合体人体检测的问题,提出一种基于梯度方向直方图(HOG)人体模型特征的检测算法。该方法通过提取人体样本库的HOG特征,用支持向量机算法(SVM)对样本的HOG特征进行分类训练。为了提高算法的精确度和适用性,以南京南站的监控视频为依据建立交通枢纽综合体人体样本库。并以南京南站监控视频和校园拍摄的人员视频作为测试集。结果证明,本算法可以有效识别交通枢纽综合体各种特征人体。

Pedestrian detection based on intelligent video of subway station

Subway platform is densely populated venues, so casualty would be disastrous if fires happen in these places. Human detection is the basis of personnel number statistics, which can guide personnel evacuation. The human detection problem in image sequences is studied in this paper, and an effective method for real-time human detection is presented. A method based on Histograms of Oriented Gradients (HOG) and human body morphological characteristics is studied to define a person. Then human samples are trained with Support Vector Machine (SVM).We shoot personal monitoring video in South Subway Station of Nanjing as the human samples. The algorithm of SVM is used to perform the training to get a human classifier, which is used to conduct the human detection. Results of experiments which were performed practically show that the proposed method can detect preceding person effectively.