摘要
针对视频序列中人脸识别效率低、扩展性差的缺陷,提出一种基于子段分割的人脸识别方法。以任务需求为牵引,将识别系统划分为前端处理子段和功能实现子段。前端处理子段实现目标行人的检测、跟踪以及特征提取;功能实现子段主要实现特征匹配及人脸识别任务。采用蝙蝠优化算法对前端处理单元进行负载均衡优化,平衡各子单元的处理时间和视频任务分配,优化CPU的利用率,降低硬件系统成本。实验结果表明,所提方法在保证人脸识别精度的同时,提升了系统的识别效率,增强了系统的扩展性能,CPU占用率改善了49.5%。
To solve the low recognition efficiency and poor scalability of the face recognition system,a face recognition method based on subsegment segmentation was proposed.The proposed system was divided into the front-end processing subsection and the function realization subsection with the task requirement as the traction.The front-end processor implemented the detection,tracking and feature extraction of the target pedestrians in the video.The function realization sub segment mainly realized the feature matching and face recognition task.The bat optimization algorithm was used to optimize the load balancing of the front end processing unit,balance the processing time of each sub unit and the assignment of video task,optimize the utilization of CPU and reduce the cost of the hardware system.Experimental results show that the proposed method improves the recognition efficiency and the system performance.The CPU occupancy rate is improved by 49.5%.
作者
王小伟
WANG Xiao-wei(Modern Education Technical Center,Physical Education College of ZhengZhou University,Zhengzhou 450052,China)
出处
《计算机工程与设计》
北大核心
2019年第10期2974-2978,共5页
Computer Engineering and Design
基金
河南省高等学校重点科研项目计划基金项目(19A520040)
郑州大学体育学院青年骨干教师基金项目(QNGGJS201804)
关键词
计算机视觉
人脸识别
子段分割
负载均衡
蝙蝠算法
computer vision
face recognition
subsegment segmentation
load balancing
bat algorithm
