Technology And Research Of 3D Measurement System Based On Linear Light Source

Since the beginning of the new century,with the continuous enhancement of my country's industrial production capacity,the requirements for measurement and machining accuracy of irregular parts in the fields of machinery manufacturing,aerospace,etc.have also been continuously improved.Therefore,the three-dimensional measurement technology ushered in the opportunity for vigorous development.Among them,the 3D measurement technology based on structured light is very popular,and market-oriented products are constantly being developed.However,the existing structured light three-dimensional measurement system has high development and use cost and large volume.Moreover,how to improve the phase unwrapping accuracy and range while ensuring time efficiency is still a difficult problem restricting the development of measurement systems.In response to these problems,this paper designs a three-dimensional measurement system device based on a linear light source,and in-depth study of the phase extraction and phase unwrapping technology based on a single frame of grating image,to achieve a low-cost,high-precision,large-range,and A structured light three-dimensional measurement system that satisfies multi-target measurement under the same field of view.The main research results of this paper are as follows:1.In view of the large volume and high cost of the traditional planar structured light based3D measurement system,this paper uses the linear light source device to replace the projector in the traditional structured light measurement system to generate the projection grating,and designs a low-cost,compact and portable 3D measurement device that can be used for a wide range of depth measurement.2.In the phase extraction process of single-frame raster images,it is easy to be interfered by the non-uniform background components.When the background components of raster images are not uniform,the existing removal methods of the projected raster background components have poor removal effect and affect the reconstruction accuracy.In order to solve this problem,we found that the background component changes very little in a single fringe period and can be treated as a constant.Combined with the characteristic that the double Hilbert transform can effectively remove the DC component of the signal and keep the phase information stable,an adaptive fractional periodic fringe background filtering method is proposed.The method firstly extracts the peak value from the projected raster image row by row and divides the fringe period one by one.Then,two Hilbert transforms are carried out to remove the background component of each fringe period,and the self-adaptive removal of the fringe background component is realized.The proposed method can effectively overcome the problem of the non-uniform background component of the fringe projection image.The experimental results show that the proposed method can better eliminate the influence of the non-uniform background component of the projection raster image and improve the measurement accuracy of the single frame raster image compared with the traditional Fourier contour technique and the empirical mode decomposition method.3.In view of the problems of error propagation and measurement depth limited in previous phase unrolling methods based on single frame image,a new high-precision and large-range phase unrolling algorithm is proposed,which can meet the requirements of multi-target measurement under the same field of view.Based on the geometric constraints between the phase of the standard reference plane and the phase of the object to be measured,the algorithm achieves depth-free 3D measurement through threshold segmentation and modulation intensity sorting method.Experimental results show that the proposed phase unwrapping algorithm has higher measurement accuracy than the existing single-frame phase unwrapping algorithm,and can simultaneously measure multiple targets.And the depth of measurement is not restricted by geometric constraints,which can accurately measure the target with large height drop.The mean square error(MSE)6.7?10^-3radof the simulation measurement accuracy is achieved.The processing time of a single frame image(effective area750piexl?750piexl)is 0.1083seconds,and the real measurement error is better than 0.15mm.