The accuracy of the detection in the image-based subgrade settlement monitoring system depend on the quality of the collected spot image, This paper studies the main factors affecting the quality of spot images and propose a method for quality assessment of spot image without reference. The method analyze the characteristics of the spot change under the influence of external conditions, and reflect the change effect of the spot feature through the size, brightness, edge definition and contrast of the spot image.Then,the four evaluation factors are integrated into one. Through the experiment, the spot images under the influence of the target surface are collected. The spot image obtained by the experiment is evaluated using the non-reference spot image evaluation method proposed in this paper, and verifies by measuring the center position of the spot image. The experimental results show that the spot image is affected by different influencing factors, this evaluation method method can combine various influencing factors to reflect the change of image quality,and can give reasonable evaluation results for spot images that are affected to varying degrees. It provides theoretical and experimental basis for the actual subgrade surface settlement monitoring system.
Aiming at the problem that the target deviation of settlement monitoring system affected the accuracy of settlement monitoring, a monitoring target attitude measurement system was designed to realize the real-time non-contact measurement of target pose. Firstly, the monocular vision and the monitoring target surface were combined, and the perspective projection model of the camera was established according to the known geometric constraint relationship between the two-dimensional feature points of the target surface. Then, for the five feature points arranged in a cross, based on the PnP problem, the HOMO algorithm was used to linearly solve the target pose. Based on this, the pose optimization solution was obtained by the nonlinear L-M algorithm. The system could perform real-time feature point image processing and pose calculation. An experimental platform for simulating the target pose measurement system in the surface settlement monitoring of ballastless track was built. The experimental results showed that the root mean square error of the three angles of the pose measurement system is 0.126°, 0.14°, 0.12°, and the whole identification process takes 1.9s. The measuring system had high calculation accuracy and was used in a short time, which was suitable for the flawless track surface settlement monitoring system.