Research And Development Of Large Surface Detection System Based On 2D Vision Sensor

Large-curved-surface parts are the cornerstone for the specific performance of aerodynamic and hydrodynamic aspects of important modern equipment such as automobiles,ships,high-speed trains,and spacecraft.The 3D information extraction method is the supporting technology for the digital design and manufacture of large curved parts.This research is derived from the project of high-speed train body machining quality inspection.A large-scale surface measurement system using 2D vision sensor is developed for automatic measurement of large curved surfaces.Aiming at surface measurement requirements and industrial environment,this paper designs the overall architecture of large curved surface measurement system based on 2D vision sensor.According to the characteristics of the measured surface,the working mode of the line structured light sensor is designed.By combining Industrial robots,linear guides and visual sensors,the system expands the detection range for large curved surfaces.The interactive software is developed for the measurement system and evaluation criterion.In order to obtain a depth image that accurately reflects the three-dimensional shape of large curved surfaces,this paper designs a depth image acquisition method based on line structure light.The 2D calibration of the camera determines its internal parameters and eliminates the image distortion.Combining the center coordinates extracted from the light bar image and the optical plane equation,the three-dimensional information of the curved surface is calculated.The depth image is enhanced by the median/bilateral combination filter,which generates an accurate surface depth image.To solve the problem that the quality criterion of the large curved surface is diverse and unable to standardize,a surface abnormal detection and evaluation method based on the curvature of the surface is proposed.The proposed method calculates the curvatures in different scales using the depth image.Then,abnormal points are detected by comparing the curvatures in small scale and relatively large scale.The difference between abnormal point and the standard is calculated by estimating the standard surface using least-square method.Finally,the system is applied to measure the high-speed train body in the industrial environment.The proposed method is compared with the original measurement method and the system shows satisfied efficiency and precision on the measurement of the large curved surface.