Automatic Centerline Extraction and Error Analysis Based on Machine Vision for Bending Pipe Image

Centerline reflects the shape characteristics of pipes, and centerline extraction is widely used in automation equipment field and Medical field. This paper proposes a method to automatically extract bending pipe's centerline with technology of image filtering and de-noising and edge detection. By finding the outer edge of the pipe, the paper creates a distance model, and the centerline of the pipe is a series of points with shortest distance between two outside edges. In error analysis, the paper analyzes the radial distortion and tangential distortion caused by a camera and also analyzes the perspective error which is very important to centerline extraction problem. By simultaneously analyzing the error of these three areas this paper presents a more accurate calculation method of the centerline. Finally, the paper verifies the correctness of the methods through experiments.