Three-dimensional Measurement Of Denture And Maxillofacial Surface Based On Line Structured Light System Research

The three-dimensional denture and maxillofacial measurement system based on line structured light is of great significance to the development of oral health.This thesis obtains different pictures multiple times from a fixed angle through the left and right cameras.First,the pictures obtained by the left and right cameras are processed.Their pixel coordinates are extracted and converted into three-dimensional coordinates which is the left and right camera point clouds.Second,the point clouds of the left and right cameras are preprocessed.Last,the point clouds of the left and right cameras are stitched,and the stitched point clouds are rendered using Open GL to complete the three-dimensional measurement of the maxillofacial surface of the isolated tooth.This thesis is mainly divided into three parts: data acquisition and processing,system calibration,and point cloud stitching.At present,the related research on system calibration mainly focuses on improving the algorithm itself,and there are few studies on the influence of external factors.Based on the above considerations,the algorithm proposed in this paper screens the obtained pixels by error screening method to reduce the influence of external factors,which not only reduces the influence of external factors,but also has simple algorithm,fast calibration speed and high calibration accuracy.1)Data acquisition and processing.The job of data acquisition is to capture images of line laser fringe that change due to the height fluctuations of the surface of the denture when line laser is projected on the surface of the measured tooth and translational movement platform drives the denture to move slowly.Data processing refers to image processing and point cloud processing.In this thesis,the image is desiccated and the center of the stripe which is sub-pixel coordinates is extracted.And the 3D point cloud obtained is simplified and desiccated.2)System calibration.The system calibration includes camera calibration and measurement system parameter calibration.By comparing the advantages and disadvantages of the linear camera model and the classic calibration method of the non-linear model,and based on the particularity of this thesis,this thesis proposed a coplanar hybrid calibration algorithm which combining error screening models,mathematical models,and neural network models based on Error Screening Model.The first step is to use the radial alignment constraint calibration algorithm based on the error screening model to solve the internal and external parameters of the camera.Due to the influence of various factors such as illumination and algorithms,there is a certain error in pixel extraction.In this thesis,inaccurate pixels are eliminated by the deviation filtering method,and the accuracy of the calibration is improved.In the second step,the camera internal and external parameters obtained in the first step areused to convert the pixel coordinates into real three-dimensional coordinates,compare the calculated three-dimensional coordinates with the actual coordinates,use machine learning to establish a compensation network and obtains compensation functions,point cloud stitching using the finally obtained 3D world coordinates.The experimental comparison proves the calibration algorithm has less error than the traditional calibration algorithm,and reduces the calibration error by about 6.5%.3)Point cloud stitching.The point cloud stitching is divided into two parts.First,the point cloud data obtained by processing the pictures obtained by the left and right cameras are stitched into two left and right point clouds.Then,the left and right point clouds are processed.And the pre-processed left and right camera point clouds are stitched by the stitching algorithm combining rough stitching of Centroid distance feature method and precise stitching of C-ICP.The final stitching results are rendered using Open GL to get the final denture and maxillofacial model.