Study On Automatic Angle Detection For Supporting Device Based On Image Processing

Supporting device mainly consist of top anchor, cantilever tube, registration arm and the steady arm, which is the main support for the high-speed railway Overhead Contact System(OCS). The posture of each supporting device component is the key factor to affect the security of OCS due to its impact on the steady arm. On one hand, if the wrong state the components stay makes the slant angle of the steady arm become smaller than before, the pantograph may hit the steady arm result from its uplift. On the other hand, if the wrong state the components stay makes the slant angle of the steady arm become bigger than before, a hard point would be produced and the abrasion increased both the contact wire and the pantograph.In addition, the posture of each component, namely, the slant angle of each component, is a very important parameter to analyze the causes of the abnormality of steady arm. Some familiar causes that make the steady arm abnormal are the follows:the abnormality of the steady arm itself, the abnormality result from the registration arm, the abnormality due to the flexible of the fastener, the abnormality result from the nonstandard installation or the slack of the support and so on. As a consequence of analyzing these causes which lead to abnormality of the steady arm, the efficiency and the security of the maintenance of the OCS is sharply improved.This paper focus on the shortage of the Catenary-Checking Video Monitor Device(CCVM) which is unable to measure the slant angles of the cantilever components in6C system. Based on the image processing of the steady arm, a method to measure the angles of each cantilever components automatically based on Hough Transform is proposed. Morphology is used more than once in order to separate the illumination and the substance in the image, what more, a few structures are produced for the segmentation between the target and the others; According to the cantilever components’characteristics, a new line segment detection algorithm is proposed, which is combined with the edge point classification technology. As a consequence, each location of the components in the image and the angles are confirmed. To achieve the matching between line segments and the components, some special rules are established. Finally, a simulation set up by a3D drawing software named Solidworks is employed to study the relationship between the real cantilever plane and the camera plane. In order to modify the error made by the different visual angle, the affine transformation is adopted. At the last of this article, a lot of image is processed by the method which is proposed in section3. Results of these images show that the line segments it detected match the real components very well and that the method proposed is feasible.