雾霭图像实时处理算法加速策略  

Real-time image processing acceleration strategy in foggy weather condition

作　　者：崔文 李强[1] 刘晓春[2] 李由[3] CUI Wen;LI Qiang;LIU Xiaochun;LI You(The PLA Unit 96941,Beijing 102208,China;College of Aeronautics and Astronautics,National University of Defense Technology,Changsha 410073,China;National Key Laboratory of Human Factors Engineering,Astronaut Research and Training Center of China,Beijing 100194,China)

机构地区：[1]中国人民解放军96941部队,北京102208 [2]国防科技大学空天科学学院,湖南长沙410073 [3]中国航天员科研训练中心人因工程重点实验室,北京100194

出　　处：《国防科技大学学报》2019年第3期106-111,共6页Journal of National University of Defense Technology

摘　　要：为仿真前视红外导引头实时匹配导引过程,完成目标实时跟踪性判断,针对限制对比度自适应直方图均衡和归一化互相关算法开展加速策略研究,提出基于统一计算设备架构(Compute Unified Device Architecture,CUDA)的雾霭图像实时处理加速方案,并进行设计参数寻优,得出线程块、线程网格设计参数存在最优尺寸,需实验测定的结论。实验结果表明,引入CUDA加速优化后,较未进行CUDA加速前时间指标提高5~20倍,能够满足目标跟踪实时性要求,可为红外导引头目标实时跟踪系统设计提供参考。In order to simulate forward looking infrared seeker′s matching guidance process,an acceleration strategy aiming at the contrast-limit adaptive histogram equalization and the normalization cross correlation algorithms was researched to realize real-time target tracking judgment. The real-time processing acceleration strategy of foggy images based on CUDA(compute unified device architecture) was proposed. The final real-time graph preprocessing acceleration schema implements two main design parameters′ optimization,including blocksize and blockgrid. Experimations show that using CUDA acceleration schema,the work efficiency is 5~20 times faster than the original algorithm. The proposed acceleration schema can satisfy the real-time target tracking judgment and can provide references to researchers devoted to the related work.

关 键 词：抗雾霭干扰 图像匹配 统一计算设备架构 加速策略 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]