| 基于智能人形识别的港口无人区监控系统关键技术 |
| 投稿时间：2013-11-19 修订日期：2014-04-03 点此下载全文 |
| 引用本文：宓为建,何鑫,沈阳,宓超.基于智能人形识别的港口无人区监控系统关键技术[J].上海海事大学学报,2015,36(1):65-69. |
| 摘要点击次数: 1713 |
| 全文下载次数: 1257 |
|
|
| 基金项目:上海市教育委员会科研创新项目（14ZZ140）；上海市教育委员会上海高校青年教师培养资助计划；上海市科学技术委员会部分地方院校能力建设专项计划（13510501800）；上海海事大学研究生学术新人培育计划（GK2013071） |
|
| 中文摘要:为改善以人工为主的传统监控系统，在天津港煤码头无人区运用智能视频监控技术进行智能人形识别监控研究.该技术利用优化的方向梯度直方图（Histogram of Oriented Gradient,HOG）算法快速对人体轮廓进行描述；结合基于港口实际场景训练得到的支持向量机分类器，标定出图像中有人的区域.在天津港煤码头无人区的现场实验表明，该技术对一幅320×240像素的图像的检测时间小于200 ms，满足港口监控实时性的要求.对具有复杂背景的监控区域，该技术能够高效地进行人形目标的匹配与识别，从而使安全得到更有效的保障. |
| 中文关键词:人形识别 方向梯度直方图（HOG） 支持向量机（SVM） 监控系统 |
| |
| Key technique of monitoring system for no man area in ports based on intelligent human profile recognition |
|
|
| Abstract:In order to improve the traditional monitoring system relying on artificial maintenance, the intelligent video surveillance technique is applied to the study on intelligent human profile recognition in no man area of a coal terminal in Tianjin Port. The technique can quickly describe the profile of human body by the optimized algorithm of Histogram of Oriented Gradient (HOG); the area where there exist some people in the images can be marked with the Support Vector Machine (SVM) classifier trained with consideration of the port real scene images. The experiment in the scene of no man area of the coal terminal in Tianjin Port shows that the processing time of a 320×240 image is within 200 ms by the technique, which meets the real time requirement of monitoring in ports. This technique can efficiently match and recognize the human profile in the monitoring area with complex background, which ensures a higher level of security and safety. |
| keywords:human profile recognition Histogram of Oriented Gradient (HOG) Support Vector Machine (SVM) monitoring system |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
