Research and Engineering Implementation of Automatic Fundus Photography Algorithm for Ultra-Wide Angle Confocal Laser Fundus Scanning System

Department of Biological Science & Medical Engineering, Southeast University, Nanjing, China

916370799@qq.com

Abstract

To achieve automatic photo capture of fundus images in an ultra-wide-angle laser-focused fundus scanning system, a pupil detection and positioning algorithm based on confocal laser technology was proposed and deployed. First, image preprocessing methods such as automatic laser intensity enhancement and contrast adjustment were applied to reduce the impact of noise and interference on subsequent processing. Next, the pupil was quickly and accurately located using the proposed pupil detection and positioning algorithm based on confocal laser technology. Finally, the movement information was calculated based on the position of the pupil center and image center, and the three-axis stepper motor was controlled in a rapid closed- loop manner to achieve three-dimensional automatic tracking of the pupil. Once the shooting conditions were met, the photo was automatically captured. To validate the effectiveness and timeliness of this solution, automatic photo capture tests were conducted on different subjects using a prototype. The success rate of automatic photo capture was 95.6%, with an average time of about 8 seconds per single-eye capture. The fundus images captured by successful automatic photos all met the requirements for automatic capture.