Wheel Tread Defect Detection Based On Photoelectric Measurement

The technical condition of the train wheel tread directly relates to the running condition of the train, and affects the safety of train operation. Due to its high efficiency and precision, photoelectric measurements gradually become a research hotspot. The limitations of traditional measuring methods that base on line-structure-light have restricted the further improvement of the efficiency and precision. Thus, it is of great application value to carry out the research of new measuring methods to improve the measurement efficiency and precision.After study on the traditional measurement model and poniting out its limitation on efficiency and precision, the paper applys the three-dimension measurement method based on fringe projection to the panoramic measurement of the tread for the first time. We established a three-dimension measurement system based on camera and projector and proposed a hardware improving program and using the three-dimension measurement model based on camera coordinate system. The measurement system collected 7 gray code images and 4 phase shifting images, then we get the relationship of the phase and the surface three-dimension system through phase shifting and resolving and system calibration. Finally we get the three-dimension model of the object through simulation and then realize the measurement of the tread parameters. The paper mainly includes:Analyze the limitations of the traditional measurement device based on line-structure-light, propose a three-dimension measurement method based on fringe projection and carry out the three-dimension panoramic measurement. By analyzing the traditional three-dimension measurement model of fringe projection, we found the existence of three strict parallel and intersect constrains. We propose an improved model based on camera coordinate system and derived the relationship between fringe phase and three-dimension coordinate.We propose an improved scheme for the hardware of the system and develop a high speed data acquisition system based on USB Protocol Control Chip 68013. We realized the hardware and software design,and then carry out rate testing experiments under different modes and optimization of the software to improve the transmission performance as well as we can.By analyzing the collected gray code images and fringe images, we approached the fringe trapezoidal distortion correction method to carry out keystone correction. By analyzing and comparing with the traditional image segmentation and filtering algorithms, we proposed the threshold segmentation method based on light and dark image and stripe image to carry out the images'threshold segmentation,and filtering the images'noise by using the spin filtering. In the work of algebra on the light and dark images and stripe images, we reduced the computation time by filtering out invalid points.We analyzed the traditional phase resolving method of fringe three-dimension measurement, with phase unwrapping, which was carried out under combination of phase shifting and gray code, finally propose a correction algorithm to correct the results of unwrapping phase and reduce phase errors.According to Zhang Zhengyou's camera calibration theory, we could set up an OPENCV environment under VC platform,and collect 11 board template image and achieve high precision calibration of the camera parameters. Besides, based on the rigid transformation theory of coordination system and the three-dimension measurement model of camera coordination system, we could achieve the parameter calibration of the phase coordinates through collecting dot array template images.