3D Body Point Cloud Data Denoising and Registration

Our team developed a portable three dimension (3D) body measurement system based on binocular stereo vision. By measuring the body, the body data of front and rear surface are reconstructed. Because of physical properties of measuring device, environment of measurement, light and the defects of calibration algorithm, the initial body point clouds include some noisy points. Based on the initial point model, the noisy points are divided into two types. The first type noisy points are far from the body data points. They are directly deleted by the distance-based algorithm. The second type noisy points mix with true body data to cause the point clouds unfair. They removed by modifying the location based on the method of mean curvature flows (MCF). The results show that the two methods are effective for denoising. By registering the corresponding markers pasted on the body, the front and the rear can be integrated as a whole body model. The registered body can be used to construct the body surface model with the sense of reality.