基于动态手势跟踪算法的视力检测系统设计

Design of visual acuity testing system based on dynamic gesture tracking algorithm

目的:设计一种基于动态手势跟踪算法的智能化视力检测系统,以满足儿童和青少年自助式视力检测的需求。方法:该系统采用基于帧间差分算法和Shi-Tomasi角点检测算法的动态手势跟踪算法实现实时识别视力检测环境中手势的移动方向,并根据视力检索算法完成相应的视力测试流程。系统硬件由Windows系统主机、显示屏、摄像头以及语音播报器等设备组成。软件遵循《标准对数视力表》的标准进行设计,并采用Python 3.7语言进行编程。结果:利用该系统进行视力检测时,手势移动识别的准确率达到97.8%,单次手势识别的平均耗时为0.568 s,双眼视力检测流程的平均耗时为99 s。经过Bland-Altman图分析,该系统采集的视力数据与通过《标准对数视力表》采集的数据之间具有高度的一致性。结论:该系统人机交互操作体验好、检测快速、准确度高,适用于家庭、学校、社区等多场景的视力检测。

Objective To design an intelligent visual acuity testing system based on a dynamic gesture tracking algorithm to meet the needs of self-administered visual acuity evaluation among children and adolescents.Methods The system realized real-time recognition of the movement direction of the gesture in the vision testing environment with a dynamic gesture tracking algorithm based on inter-frame difference method and Shi-Tomasi corner detection algorithm,and completed corresponding vision testing processes with the vision retrieval algorithm.The hardware of the system was composed of a Windows system host,a display screen,a camera and a voice announcer,and the software was designed according to "Standard Logarithmic Visual Acuity Chart" and programmed with Python 3.7 programming language.Results When used for visual acuity testing the system developed had the accuracy being 97.8% for gesture movement recognition,the average time consumed being0.568 s for one times of gesture recognition and 99 s for binocular vision testing.Bland-Altman plots showed the system developed had high consistency with "Standard Logarithmic Visual Acuity Chart" in captured visual acuity data.Conclusion The system developed gains advantages in human-machine interaction,rapid testing and high accuracy,and can be used for visual acuity testing in multiple scenarios such as home,school and community.