基于网格化特征提取的恐高症筛查方法
Acrophobia screening method based on gridding feature extraction
刘帅 1鲍本坤 1成贤锴1
作者信息
- 1. 中国科学技术大学生物医学工程学院(苏州),生命科学与医学部,苏州 215163;中国科学院苏州生物医学工程技术研究所,苏州 215163
- 折叠
摘要
基于视觉探索特征的恐高症筛查方法主要运用瞳孔特征,然而瞳孔特征易受环境光强干扰,严重影响恐高症筛查模型准确性.为解决该问题,提出了基于网格化的视觉探索特征提取算法,对视觉探索数据进行网格划分,解耦光强信息后再提取相关信号特征和行为特征;同时提出了主动视觉探索诱发范式来更好地诱发恐高状态.实验结果显示:基于网格化的视觉探索特征提取算法准确率提高了 11.77％,同时主动视觉探索诱发范式的恐高诱发程度提高了 13.57％.结果表明,基于网格化的视觉探索特征提取算法和主动视觉探索诱发范式具有较好的应用前景.
Abstract
The screening methods based on visual exploration features mainly use pupil features.However,pupil features are easily disturbed by ambient light intensity,which seriously affects the accuracy of the acrophobia screening model.To solve this problem,the grid-based visual exploration feature extraction(GVEFE)algorithm is proposed.This method involves dividing visual exploration data into grids and decoupling light intensity infor-mation before extracting relevant signal and behavioral features.Meanwhile,the active visual exploration elicita-tion(AVEE)paradigm is also proposed to better induce the acrophobia state.Experimental results show that the GVEFE improves the accuracy by 11.77％,while the AVEE paradigm improves the degree of acrophobia elicita-tion by 13.57％.The results show that the GVEFE algorithm and the AVEE paradigm hold promising application prospects.
关键词
网格化/视觉探索/特征提取/虚拟现实/恐高症Key words
gridding/visual exploration/feature extraction/virtual reality/acrophobia引用本文复制引用
出版年
2024
NETL
NSTL
万方数据