A Visual Monitoring Method for Monochromators Based on Video Super-Resolution and Kalman Filter

• Published in: 2024 5th International Conference on Intelligent Computing and Human-Computer Interaction (ICHCI) pp. 286 - 290
• Main Authors: Li, Zhijuan, Ma, Shiwei
• Format: Conference Proceeding
• Language: English
• Published: IEEE 27.09.2024
• Subjects: Filtering algorithms; Human computer interaction; kalman filter; Kalman filters; Machine vision; Monitoring; monochromator; Real-time systems; Streaming media; Superresolution; video super-resolution; visual monitoring; Visualization; YOLO
• DOI: 10.1109/ICHCI63580.2024.10807971

To achieve real-time and precise monitoring of the main moving component of the monochromator. We have developed a visual monitoring system for monochromator based on machine vision technology for SSRF. To enhance the accuracy and stability of visual detection, this paper proposes a visual monitoring method that integrates video super-resolution (VSR) and kalman filter (KF) with YOLOv5 object detection. First, VSR is applied to enhance the quality of images captured by the camera. Then, scale numbers on a ruler are detected, and KF is employed for stable tracking of the scale numbers' detection boxes. Finally, the position of the monitored subjects is calculated based on tracking results of the scale numbers on the ruler. Experimental results demonstrate that by implementing VSR, the detection rate of scale numbers improved by an average of 12%, and in combination with the tracking algorithm based on KF, the proposed method achieves enhanced stability in the positioning of the moving component in the monochromator.