Scholarlink

This study introduces an automated system for detecting strabismus and analyzing corneal deviations using advanced image processing techniques. The system classifies different types of squints by calculating the Vertical Equation Ratio (VER) and Horizontal Equation Ratio (HER). Additionally, corneal deviations are quantified through the Cosine Angle Method to determine Horizontal and Vertical Corneal Degree Deviations (HCDD and VCDD). A vector-based direction analysis is utilized to identify the direction of deviation, with positive values indicating exotropia or hypertropia, and negative values indicating esotropia or hypotropia. The methodology integrates Mediapipe that was developed by Google with large datasets for precise feature point detection utilizing 468 facial landmarks for precise eye and iris boundary detection and applies geometric and trigonometric computations for enhanced accuracy. The proposed algorithms effectively classify squint types—esotropia, exotropia, hypertropia, and hypotropia—and measure their respective degrees of deviation.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

AI-Enabled Automated Strabismus Detection and Corneal Deviation Analysis Using Image Processing Techniques

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Govardhana Rekam,
  • Govind N. Sahu

摘要

This study introduces an automated system for detecting strabismus and analyzing corneal deviations using advanced image processing techniques. The system classifies different types of squints by calculating the Vertical Equation Ratio (VER) and Horizontal Equation Ratio (HER). Additionally, corneal deviations are quantified through the Cosine Angle Method to determine Horizontal and Vertical Corneal Degree Deviations (HCDD and VCDD). A vector-based direction analysis is utilized to identify the direction of deviation, with positive values indicating exotropia or hypertropia, and negative values indicating esotropia or hypotropia. The methodology integrates Mediapipe that was developed by Google with large datasets for precise feature point detection utilizing 468 facial landmarks for precise eye and iris boundary detection and applies geometric and trigonometric computations for enhanced accuracy. The proposed algorithms effectively classify squint types—esotropia, exotropia, hypertropia, and hypotropia—and measure their respective degrees of deviation.