Research On 3D Human Body Measurement Technology Based On Multi-direction Point Cloud Piecing

In recent years,anthropometric techniques have gradually got rid of the constraints of traditional manual measurement methods,and gradually combined with computer technology to form a three-dimensional body measurement technology.The 3D anthropometric measurement technology has the characteristics of fast and accurate measurement,suitable for a large number of human body data acquisition,suitable for commercial scene applications,and can provide effective and practical data in personalized clothing customization,virtual fitting,and ergonomics research.Among the existing three-dimensional anthropometric techniques,commonly used measurement techniques include laser measurement,structured light scanning,and moiré measurement methods.The instruments used in these measurement methods are complicated,cumbersome,and expensive,and cannot be adapted to daily needs.The purpose of this paper is to propose an anthropometric measurement program that is efficient,low-cost and suitable for personal use.It abandons the threedimensional reconstruction of the cumbersome human body model,and uses the combination of multi-directional point cloud data to extract the human body girth data of interest..The main research contents of this paper are:(1)3D point cloud optimization based on corrosion algorithm.Because Kinect reduces the accuracy of point cloud data when capturing the edge of the object,this paper needs to remove the captured point cloud data.However,there is no corrosion algorithm for 3D point cloud in the existing algorithm.Therefore,this paper proposes a three-dimensional point cloud corrosion scheme based on two-dimensional image erosion algorithm,which maps the corrosion mode of the point cloud two-dimensional contour to the three-dimensional point cloud,and completes the edge removal of the three-dimensional point cloud to optimize the three-dimensional point cloud.(2)Point cloud edge extraction method based on grid filtering.The edge of the cloud is an important basis for the integration of the human point cloud.However,the traditional point cloud edge extraction method is very efficient in the face of a large number of point cloud data,which cannot guarantee the efficiency of the scheme.Therefore,edge extraction scheme is improved,and a fast point cloud model edge acquisition method based on grid filtering is proposed.(3)Human body point cloud segmentation technology based on point cloud 2D contour and bone point data.Since the small limb rotation of the measuring object affects the integration of the human point cloud,it is necessary to divide the human point cloud and then combine the different parts.The traditional human body model segmentation scheme usually only uses the contour features of the human body for segmentation.In addition to utilizing the contour features of the human body point cloud,this paper makes full use of the bone point data to divide the human body into multiple parts,and the segmented point cloud is excluded.They interfere with each other and completely preserve the three-dimensional shape of each part.(4)Human body point cloud stitching technology based on point cloud 2D contour and bone point data.In order to obtain complete human body circumference data,it is necessary to combine multiple pieces of point clouds of the human body.The traditional point cloud registration algorithm involves a large number of iterations,which leads to time consuming.In this paper,for the characteristics of anthropometric measurements,the universal registration algorithm is abandoned,and the features of the human contour and the positional features of the bone points are used as the key points for the flattening,so that the flattening process can be completed in only a few seconds.(5)Segmented polynomial curve fitting technique.Because the discrete points lack the description of the human body contour,curve fitting is needed.In the traditional curve fitting scheme,the single-curve curve is often used to fit the girth data,which cannot cope with the complex situation such as curve shape of the female chest curve.Therefore,this paper proposes a curve fitting technique that first divides the girth data into two groups,and then performs polynomial fitting separately.Polynomial can express the function curve of any shape,so it is a good solution to the curve fitting problem of complex contours.Finally,this paper tests and verifies a large number of measurement objects,which proves that the anthropometric measurement scheme is feasible and effective,and the accuracy is high.