Research On Active Data Fusion Technology Based On Line Structure Light

With the rapid development of China's manufacturing industry and the continuous progress of computer technology,the machine vision technology represented by non-contact measurement has gradually become a research hotspot.Among them,non-contact measuring system based on linear structured light sensor has become one of the most potential and widely studied technologies due to its advantages of large range,high precision and high stability.Traditional line structured light sensor is composed of a single camera and a laser(LD)light source,there are some of these limitations: first of all,the single camera on surface topography more complex when the measured object to be tested,will be unable to get all the surface topography data due to shade,and the single camera to capture multiple surface of the object to be tested at the same time,could not guarantee the accuracy of measurement.Secondly,the surface of the object irradiated by laser presents granular structure,namely laser speckle.The existence of speckle will reduce the measurement accuracy.In view of the above problems,this paper adds a camera on the basis of the single-camera wire-structured light sensor,and uses LED instead of laser as the light source of wire-structured light,and designs a non-contact measurement system based on the dual-camera wire-structured light sensor.The research work of this paper mainly includes the following aspects:1.The measurement model of linear structured light measurement system is analyzed,and the measurement principle of linear structured light is deeply studied.According to the projection model to complete the system hardware selection,system construction.2.Study the line structured light measurement system calibration involved,and completed the camera parameter calibration based on Zhang zhengyou's calibration method.,on the basis of camera calibration,the calibration of optical plane and direction cosine is completed,the pose of optical plane and guide rail under the camera is acquired,and the calibration between two cameras is completed based on two-dimensional plane target,and the rotation matrix and translation vector between cameras are acquired,laying a foundation for subsequent data fusion.3.The structure light sensor driven by the guide rail can scan the object under test.The two-dimensional image is preprocessed based on the OpenCV library.The precise pixel coordinates(u,v)of the center point are obtained by using the improved gray gravity method.Through to obtain the parameters of the camera and optical plane calibration transfer two-dimensional pixel coordinates for the two cameras(corresponding to different surface of the object under test)of the 3D point cloud,completed by calibration to obtain the parameters of the direction cosine of point cloud data under the same camera,according to the double camera calibration to obtain the parameters of the complete data fusion between the camera and eventually get the object to be tested the whole 3D point cloud,and extract the geometric parameters of interest.The 3D topography of the spliced and fused point cloud was reconstructed,and the geometric parameters of the measured object were extracted and compared with the standard values.The practicability and accuracy of the system in this paper were verified,and higher accuracy was obtained.