一种高可靠性的头部校验夜间行人快速识别方案

针对夜间行人检测成像尺度不一等因素所引起的类内方差较大、实时性不足等问题，本文在统计学习的应用原理下，设计了基于熵加权和FCSVM优化的头部校验夜间行人快速识别方案.该方案应用熵加权原理改进梯度直方图特征，引入了三分支结构的支持向量机对目标进一步识别，同时利用快速分类支持向量机(FCSVM)降低运算所需的开销，确保实时性，最后通过头部校验方法对误检目标进一步分析评估，进一步提高图像匹配的准确度.实验结果表明，该方案在夜间环境下能有效区分远红外行人目标，在充分确保行人实时性的基础上，在市区、郊区等不同应用环境中，均具有良好的实用性.

A rapid and high reliable dentify program for nighttime pedestrians

For the larger intra class variance and inadequate real time problems caused by factors of imaging scales difference in nighttime pedestrian detection. The paper designs a rapid dentify program for nighttime pedestrians based on entropy weight and header checksum of FCSVM optimization under the the application of the statistical learning principles. The program utilizes entropy weighted to improve the feature of gradient histogram, introduces three branch structure SVM to identify the target further,and uses rapid classification FCSVM to reduce the overhead required of computational and to ensure real time, finally through the header checksum method to analysis and assess error detection goals, to further improve the accuracy of image matching. Experimental results show that the scheme can distinguish far infrared pedestrian goals effectively at night environment, and have good usability in urban,suburban and other different application environments on the basis of ensuring pedestrian real time fully.