用户登录
期刊信息
  • 主管单位:
  • 中国科学技术协会
  • 主办单位:
  • 中国仪器仪表学会、上海光学仪器研究所、中国光学学会工程光学专业委员会
  • 主  编:
  • 庄松林
  • 地  址:
  • 上海市军工路516号上海理工大学《光学仪器》编辑部
  • 邮政编码:
  • 200093
  • 联系电话:
  • 021-55270110
  • 电子邮件:
  • gxyq@usst.edu.cn
  • 国际标准刊号:
  • 1005-5630
  • 国内统一刊号:
  • 31-1504/TH
  • 邮发代号:
  • 单  价:
  • 15.00
  • 定  价:
  • 90.00
漂移算法在视敏度检查中的应用
Research on vision testing model based on mean shift algorithm
投稿时间:2024-06-09  
DOI:10.3969/j.issn.1005-5630.202406090076
中文关键词:  概率分布  数学期望  均值漂移算法  视敏度
英文关键词:probability distribution  mathematical expectation  mean shift algorithm  visual acuity
基金项目:国家自然科学基金(61605114,52206102)
作者单位E-mail
刘建廷 上海理工大学 健康科学与工程学院,上海 200093  
项华中 上海理工大学 健康科学与工程学院,上海 200093 xiang3845242@usst.edu.cn 
马乐飞 上海理工大学 健康科学与工程学院,上海 200093  
程慧 上海理工大学 健康科学与工程学院,上海 200093  
王冰城 上海理工大学 光电信息与计算机工程学院,上海 200093  
摘要点击次数: 790
全文下载次数: 372
中文摘要:
      为了实现对受测者视敏度的精确测量,建立了一种结合概率分布与均值漂移算法的视敏度检测系统。该系统利用均值漂移算法沿数据密度递增方向探索聚类中心的特性,结合概率分布理论,精确计算不同视敏度水平对应的视敏度概率。通过反复迭代,不断更新探索点并使其向概率密度较高的区域移动,直至找到最符合受测者实际视敏度水平的点。在系统实现过程中,算法可根据受测者响应实时调整视标的尺寸,通过持续地将当前观测值向其数学期望或均值方向调整,经过多轮迭代使概率分布逐渐贴近实际视敏度分布。实验结果表明,在3个实验场景中,即使在错误发生频率最高的情况下,该算法的平均误差率也仅为0.0083,基本满足视敏度测量的稳定可靠、鲁棒性高、抗干扰能力强等要求。
英文摘要:
      To achieve precise measurement of visual acuity of testers, a visual acuity detection system that combines probability distribution and mean shift algorithm has been developed. This system integrates the probability distribution with the mean shift algorithm, utilizing the characteristic of the mean shift algorithm, which is to explore the cluster centers along the direction of increasing data density, in conjunction with probability distribution theory, to accurately calculate the visual acuity probability corresponding to different levels of visual acuity. Through repeated iterations, the algorithm continuously updates the exploration points, moving them towards areas of higher probability density until finding the point that most closely matches the actual level of the subject's visual acuity. In the implementation process, the algorithm adjusts the size of the visual markers in real-time based on the responses of the subjects. By continuously adjusting the observations towards their mathematical expectations or means, and through multiple rounds of iterations, the probability distribution gradually approximates the actual distribution of visual acuity. Experimental results show that, across three experimental scenarios, even in the case with the highest frequency of errors, the average error rate of the algorithm is only 0.0083. This essentially meets the requirements for stable and reliable visual acuity measurement, high robustness, and strong anti-interference capability.
HTML   查看全文  查看/发表评论  下载PDF阅读器
关闭