Breast Three-Dimensional Reconstruction and Feature Points Tracking

In recent years, three-dimensional human body reconstruction technology has been widely applied and developed, and three-dimensional reconstruction can comprehensively and accurately observe the movement patterns from a three-dimensional perspective. At this stage, 3D human body reconstruction technology is mostly applied to static whole-body measurements, and there are fewer studies on breast local modelling. Moreover, the special structure of the breast area is that the breast does not follow the trajectory of the torso completely during the movement, and will be displaced relative to the torso. This study proposes a measurement method for tracking feature points based on three-dimensional breast dynamic reconstruction. In this paper, the Kinect body sensor is used to capture dynamic images of the breast, and the raw point cloud is preprocessed to generate a coloured point cloud set of the breast. The initial alignment is performed using the SMPL model, and then the ICP alignment algorithm is used for accurate alignment, while the vertex displacement D component is optimised to fit the 3D imaging and generate the final 3D breast dynamic model. Feature points are marked on the breast model, and the displacement parameters of the breast relative to the torso in the motion state are obtained by tracking the trajectory of the feature points. This method constructs a more accurate breast model and better describes the breast motion mechanism.