中文摘要：

对机场区域中人群涉暴恐动作进行及时准确的识别对于维护机场安全稳定具有重要的现实意义。传统的方法都是采用固定场景下的特定特征的人员识别,由于没有考虑机场区域的非限定性和暴恐人员的伪装性,使得识别问题变成高维的问题,降低了人群暴恐动作识别的准确性和及时性。提出一种基于投影特征算法的机场区域中人群涉暴恐动作智能识别方法。对采集的原始监控图像进行特征重构,在重构过程中选取最优代价函数,为涉暴恐动作的准确识别提供准确的数据基础。将重构结果二维图像在一维平面进行投影,转化为一维识别问题,综合递推关系对涉暴恐动作进行分类识别,降低了识别时间,提高了识别的准确性。实验结果表明,利用改进算法进行机场区域人群涉暴恐动作识别,能够提高识别的准确性和及时性。

英文摘要：

Timely and accurate identification for critical fear action of people involved in airport areas has impor- tant practical significance for maintaining airport security. The paper put forward an intelligent recognition method based on projection feature algorithm for the crowd involved in violence action at airport area. We collected the char- acteristics of original monitoring image, selected the optimal cost function in the process of reconstruction, and pro- vided accurate data base for accurate identification of involved violence fear actions. The reconstruction results of 2 - d image was projected to a one - dimensional plane, and transform it into one dimensional recognition problem.. The recursive relations were integrated to classify the involved violence fear actions and recognize them respectively, whch can reduce the recognition time and improve the precision of identification. The experimental results show that, the algorithm can improve the recognition accuracy and timeliness.