Research On Profilometry Extraction And Mosaic Technology Of 3D Grating Projection

Grating projection 3D data measurement technology has been widely used in Ar / VR,consumer electronics,industrial production control,cultural relics protection,biomedicine and other fields,with increasing demand.High quality point cloud model can be obtained by structured light projection.In this process,the effective area of the object needs to be extracted,and the extraction results directly affect the final reconstruction shape of the point cloud model.On the other hand,when obtaining the complete point cloud model of an object,it is also necessary to register the point cloud which is not in the same coordinate system from different perspectives,and the registration results have a great impact on the reconstruction results of the model.Aiming at the problem of effective region extraction and point cloud data registration in object point cloud model,this paper proposes improved methods respectively.The research work of this paper mainly includes:(1)The measurement principle and calibration technology of three-dimensional grating projection measurement system are studied.This paper focuses on the three-dimensional measurement principle,camera calibration principle and the calibration principle of solving the phase-3d coordinate mapping relationship based on the phase-shifting method.On the calibration system,the four-step phase-shifting algorithm is used to complete the three-dimensional measurement.(2)In order to improve the extraction speed and efficiency of the effective area of the measured object,an improved k-means clustering algorithm is proposed to automatically extract the effective points of the contour of the measured object,and Otsu algorithm is used to optimize the classification of the effective points to further increase the effective points of the contour.First of all,the fringe image of the measured object is obtained by using the structured light projection phase-shifting method,and the modulation image is calculated.According to the principle that the modulation values of the measured object and the background are different,the points of the modulation image are divided into the measured object point,the background point and the measured object boundary point by using the improved k-means clustering algorithm.The improved k-means clustering algorithm belongs to the hard allocation algorithm.There is a number of effective points of the measured object in the background points of the allocation results.In this paper,Otsu algorithm is used to classify the background points twice,and finally the effective points of the measured object are obtained.(3)In order to improve the accuracy of point cloud registration,a weighted improved ICP algorithm based on adaptive overlap estimation is proposed.The closer the ratio of the forward distance to the back distance of the point pair of the source cloud is to 1,the greater the probability that the point pair is the correct response point pair.Generally,the forward distance is greater than the back distance.Using this feature,setting a threshold can distinguish the point cloud in the overlapping area from that in the non overlapping area.In order to solve this problem,an adaptive threshold is proposed to divide the overlapped area.In order to make full use of the characteristics of the point-to-distance ratio,the point-to-distance ratio and the exponential function are combined to complete the point cloud registration.In the process of point cloud registration,the closer the two-way distance ratio is to 1,the greater the weight value is given to the corresponding point pair,otherwise,the smaller the weight value is given.In each iteration of registration,update the two-way distance and weight of point pair and point pair,and cycle the above steps to complete point cloud registration.On the basis of the three-dimensional grating projection measurement system,the two algorithms proposed in this paper are verified.The experimental results show that the algorithm proposed in this paper can extract the effective points of the measured object efficiently and quickly,and the improved ICP algorithm can realize the accurate registration of the point cloud.