Flatness Measurement of Truck Rack Girder Based on Machine Vision

摘要

According to the characteristics of truck rack girder and testing requirements on site in the workshop, proposes an online detection scheme based on the ROI (region of interest) and sequential images as well as improving stripe center extraction method which is on the basis of the grayscale barycenter algorithm. It is proved that commonly used Steger algorithm takes 393ms to extract stripe center of one frame image while the paper algorithm just needs 5ms of which the extraction precision reaches sub pixel. In order to decrease measurement error generated by the manufacturing technology level and mechanical vibration, proposes a global calibration method based on multiple calibration plates. By flatness measurement of 1.2 meter truck rack girder, it is concluded that the absolute error is 0.12mm between the method of this paper and three coordinate measuring machine. It can meet steadily the requirements of online detection.