The Key Technology Of Vision Measurement And It’s Application In The Railway Ballast Parameter Measurement

Vision measurement technology is a new measurement technology based on computer vision, and combines with maths, optoelectronics, image processing technology and other subjects. It is one of the core technology of modern detection technologies and one of the hot research in the field of computer vision. In recent years, with the rapid development of China Railway,vision measurement technology with its rapidity, simpleness, non-contact and other advantages in a safe condition monitoring has been widely applied.The particle shape of ballast, which is a major component of ballasted track bed, directly affects the service performance of ballast bed.So carrying out an accurate, efficient, convenient ballast geometry parameter measurement method is of great significance for maintaining the security of railway line.Based on the key technology of vision measurement, we propose a ballast double contour extraction algorithm based on dual-plane mirror camera. And for the geometry of the characteristics of ballast particles, we improve the camera intrinsic parameters of self-calibration algorithm, as a basis, complete the ballast geometry parameter measurement methods.The main contributions of this thesis are as follows:(1)We proposed a multi-view ballast contour extraction algorithm based on double plane mirrors system. Aiming at the mirror image and the particularity of ballast color priors, we estract the multi-view contour of ballast by the steps of region segmentation, local binarization, denoising and smoothing, and we also analyze the error tolerance. The ballast measuring experiment proves that the algorithm is effective and practical.(2)We put forward an improved internal calibration method based on double plane mirrors system. First we use the multi-view contour of ballast from the mirrors to complete the camera self-calibration, aiming at the experiment object ballast’s nonconvex character. So the improvement of the internal calibration method base on calibration target is proposed. The experimental results show that the improved calibration method is effective.(3) We proposed the measurement method of ballast parameter based on a two-dimensional multi-view contour and3D geometry reconstruction. According to the contour projection to the multi-view outline of ballast, we study the geometric relationship between the ballast contour and plane mirror imaging, and present the calculation method of ballast parameters based on two-dimensional contour. We also use the three-dimensional data of the ballast to complete the ballast parameter measurement. The experimental results show the effectiveness of the measurement method.