Computer Engineering and Applications ›› 2021, Vol. 57 ›› Issue (23): 37-49.DOI: 10.3778/j.issn.1002-8331.2107-0279

• Research Hotspots and Reviews • Previous Articles     Next Articles

Survey on ERP-Based Target Detection

LI Zixing, LAN Zhen, TANG Dengqing, YAN Chao, XIANG Xiaojia, ZHOU Han   

  1. College of Intelligence Science and Technology, National University of Defense Technology, Changsha 410073, China
  • Online:2021-12-01 Published:2021-12-02

基于ERP信号的目标检测研究综述

李子杏,兰珍,唐邓清,闫超,相晓嘉,周晗   

  1. 国防科技大学 智能科学学院,长沙 410073

Abstract:

Target detection technology is widely used. The existing target detection methods based on computer vision are difficult to meet the needs of complex scenes due to target occlusion, light intensity and other factors. The advanced cognitive ability and rapid perception ability of the human brain have certain advantages in dealing with complex situations. Event-Related Potentials(ERP) based Brain Computer Interface(BCI) system is related to specific events. This system can detect high-level cognitive activities independent of spontaneous EEG. It has become one of the hotspots of artificial intelligence research. In this paper, the research status of target detection based on ERP signal is summarized. First, the experimental paradigm based on Rapid Serial Visual Presentation task(RSVP) is introduced, including the design factors such as presentation mode, target field of view, target complexity and so on. Then, the preprocessing methods, feature extraction and feature classification algorithms in EEG analysis are summarized, and the applications in face recognition, military operations, medical analysis and other fields are introduced. Finally, the problems and challenges in the current research are discussed, and the future research direction and application prospects are prospected.

Key words: Brain Computer Interface(BCI), Event-Related Potentials(ERP), target detection, Rapid Serial Visual Presentation(RSVP), signal analysis

摘要:

目标检测技术应用广泛,现有的基于计算机视觉的目标检测方法由于目标遮挡、光照强弱等因素难以适应复杂场景的需求。而人脑的高级认知能力和快速感知能力在处理复杂情况时具有一定的优势。基于事件相关电位(Event-Related Potentials,ERP)的脑机接口(Brain Computer Interface,BCI)系统与特定事件相关,可检测独立于自发脑电的高级认知活动,是当前人工智能领域的研究热点之一。针对基于ERP信号的目标检测各个环节进行了研究现状的全面归纳,梳理了以快速串行视觉呈现(Rapid Serial Visual Presentation task,RSVP)为主的实验范式,包括呈现模式、目标视场角、目标复杂度等设计因素。总结了脑电信号分析中的预处理方法、特征提取和特征分类算法,介绍了其在人脸识别、军事作战、医学分析等领域中的初步应用。探讨了目前研究中存在的问题和挑战并展望未来的研究方向与应用前景。

关键词: 脑机接口(BCI), 事件相关电位(ERP), 目标检测, 快速串行视觉呈现(RSVP), 信号分析