Visual Feature Recognition of an X-ray Based Inspection Pipeline Robot

Abstract

An X-ray based pipeline robot is developed in order to inspect the welding of pipeline. The robot consists of the inner robot which carries X-ray device and searches the welding seam, and the outer robot which capture X-ray image via an enhancer. Based on visual recognition of identified lead wire, the paper presents a synchronous motion control method, simplifies the Prewitt operator to detect the edge image quickly and recognize vertical or horizon edge. A hump-shape recognition algorithm is studied to recognize the welding seam and the labelling lead wire. The experiments show that the simplified Prewitt edge detection approach and the gray-scale hump-shape recognition algorithm is valid and efficient, and the visual feature recognition algorithm has a satisfied application in pipeline robot.