An Area Weighted Surface Sampling Method for 3D Model Retrieval

3D model feature extraction is a key step for geometric content based 3D model retrieval. A 3D model is usually expressed by patches and it has large amount of data. Pseudo-random sequential Monte Carlo surface sampling method is used to sample 3D model data. It has strong randomness. The sampling results are not uniform, which worsen the efficiency and precision of feature extraction. An area weighted surface sampling method for 3DModels is proposed. The larger area patches get the higher probability of sampling. The sampling points are evenly distributed in sampling space, which represent more accurate 3D model features. Experiments show that the proposed method needs less computing resources. Even when there are less sampling points, it can still keep high calculation accuracy and stability.