Research on internal defect detection method based on machine vision

The narrow internal space, inadequate lighting, and complex geometry of internally threaded pipes pose significant challenges for accurate detection. This paper presents a computer vision-based solution that enhances image acquisition efficiency through innovative hardware design. By utilizing cylindrical stitching algorithms, we achieved panoramic image stitching of internal threads, and automated defect detection was implemented using a YOLO-based object recognition algorithm. This research addresses specific industrial needs for internal thread detection in oil pipes. Compared to existing technologies, this method significantly increases detection speed, reduces the risk of human error, and improves production efficiency.