Inspection Of Cable Surface Defects Design By Image Processing

Cable-stayed bridge is considered the main type of large span bridges in China currently. The bridge deck is connected with the tower by a lot of wire ropes and the tower consists of the cables tower, the main girder and stayed-cables. As major parts of the cable-stayed bridge, the cables are also major load-carrying parts. As the cables are exposed to the outside and eroded by something like the sunlight and rain for a long time, the surface of them is also subjected to erosion, which will inevitably lead to the erosion of the inner wire ropes, thus affecting their carrying capacity. Therefore, it is very necessary to detect the defects on the surface of the cables. Due to the great length of the cables and distance from the ground, manual testing will be very dangerous and strenuous, so there is necessary to design an automatic detection system for such defects.The automatic detection system for defects of cable surface includes three parts: the cable climbing robot, DM642-based image acquisition and the transmission system, cables surface image processing system. In this paper, a series of arithmetic are designed to do defect images segmentation and features extraction according to the actual situations and characteristics of the cables surface images.The defects on cable surface are complex, and the arc shape makes the uneven of sunshine illumination increase. So it is very difficult to design the algorithm that accurately extracts those defects. Cable surface defects extraction includes cable edge detection, cable surface defect segmentation and feature extraction.The objective of cable edge extraction is to remove the interference in background of the image, so as to prepare for the later cable surface defects extraction. The canny algorithm is selected after analyzing some routine edge extraction algorithm, aiming at the characteristics of cables edge, to process the cables edge in edge images based on the HOUGH change and pick up the cables surface area of the images.Cable surface defect image segmentation is the key part of this paper. Because of the complex situation of the cable surface images, it is difficult to select the common image extractions algorithm. The cable surface defeats are categorized into smooth and rough defeats after analyzing the characteristics of the cable surface images, and two algorithms are designed to the different types of defeat. Then, based on shape characteristics and intensity characteristics of the defect classification, the features are extracted respectively.The focus of this paper is the process of cable surface defeats extraction, which is also the main content of my paper. After repeated argument, the proposed algorithm is good applicability to solve the surface defects extraction on the cable automatically and a reference to some kind of image processing.