Research On Non-Contact Human Body Size Measurement And 3D Modeling Based On Vision Technology

Human body measurements have a wide range of applications in fields such as clothing customization,sports science,and medical applications.Traditional human body measurement techniques require the use of external tools and manual contact-based measurement.However,using a scanner for human body measurement has the disadvantages of long measurement time and high cost.This article proposes a non-contact human body measurement method based on visual technology,and designs and develops a 3D human body model generation system that can freely rotate,scale,and resemble the user’s body shape.This article mainly studies the following two key issues in the non-contact human body measurement process: The first issue is that after capturing the human body image,due to the interference factors such as complex background,clothing,and lighting in the image,the human body edges and background areas may be correlated and cannot be well segmented,which affects the extraction of the human body contour.The second issue is that the current algorithms for edge detection and corner detection in images are mostly based on curvature calculation,corner models,grayscale changes,and other methods,each with their own limitations in human body measurement applications.This article proposes corresponding optimization and improvement methods for the above two issues.To address the first issue,this article proposes an improved OTSU segmentation method based on the maximum inter-class variance method(OTSU segmentation method).In determining the segmentation threshold,the method introduces weights to obtain the optimal segmentation threshold,thus achieving a complete human body region.This method can separate the background image from the target,thereby obtaining more accurate and complete human body contours in complex environments,effectively reducing the negative impact of factors such as background and clothing on human body contour extraction.To address the second issue,this article designs a corner detection algorithm based on the multi-scale maximum chord length.First,the edges extracted from the image are convolved using Gaussian filters at three different scales,and then the candidate corner points are extracted using the maximum chord length method on the convolved edges.Then,the non-maximum suppression method is used to remove false corner points,and the true corner points are finally output.This algorithm greatly improves the accuracy of detecting and selecting characteristic points in various parts of the human body.In order to verify the effectiveness of the proposed algorithms,corresponding experimental tests were conducted.The experimental results showed that the improved corner detection algorithm achieved at least a 35% increase in accuracy compared to the CSS algorithm(Curvature Scale Space,CSS),and the image segmentation effect of the improved OTSU method was also excellent.The human body measurement experiment results showed that the average error of the two-dimensional size of the measured human body in this article was 1.47 cm,and the average error of the three-dimensional size was 2.69 cm,which meets the application requirements of human body measurement.