Feature-based Mesh Processing

The development and progress of time, science and technology made the multimedia technology integrated into every corner of our lives quickly. In the fields of computer graphics, computer animation, geometric modeling et al. The object is usually described by a 3D mesh model. It can not only fully restore the characteristics of the object itself, but also benefit for the hardware of computer graphics to process data directly. For this reason, 3D mesh model is widely used, and in the same time, the number of the models is increasing dramatically, we have to face the huge burden of computer storage space, processor processing efficiency, network transmission and fast retrieval et al. caused by such large model data. The problem of simplification and retrieval for the 3D mesh model is imminently.As the important tasks in mesh model processing, the simplification and similarity measurement have been in-depth study and research by experts and scholars which in related fields, achieved a lot of stage results. A variety of simplification methods and similarity measurement algorithms emerge in endlessly and development. The simplification of mesh model is not simply to reduce the number of triangles in the model. Its essence is that As far as possible with the least vertices and patches to represent the original model on the basis of the appearance and geometry features of the original model are preserved to the maximum extent. This requires us to delete the unrelated patches of the features for model as much as possible under the circumstances that not affect the accuracy of simplified model. The key problem of simplification the models is how to choose these patches. The retrieval technique of 3D mesh model is mostly used to provide the function of searching mesh models quickly which meets the requirements of the user. Like image, audio, video and other data retrieval, the two most basic requirements of the search are the running speed and the accuracy of the retrieval. For the simplification of 3D mesh model retrieval, its core technology is to extract what kind of feature elements to do similarity matching. In other words, it means that which feature can represent the difference between original model and other models. The selection of feature can affect the efficiency and precision of the later retrieval directly.In the aspect of mesh simplification, aim to the problem that how to increase the fidelity of simplified model, this paper put forward a method which using vertex clustering based on principal curvature. Our measure taking the gradual simplified thinking and simplifying the mesh model successive by using bounding box to subdivide it, quadric error metric and principal curvature based on integral invariants. The experimental result shows that our method not only improve the defect of error accumulation which consisted in the previous algorithms, but also preserving the detail characteristics and ensuring the display effect for the model, while raising the efficiency of algorithm, reducing both time complexity and space complexity.According to the problem that which features we can choose and how to extract the feature for the similarity measurement in 3D mesh model, this paper put forward to make sure the matching areas by structuring and comparing Octrees of the point cloud data for the areas which will be matched. We use the integral invariants to estimate the principal curvature direction of points which selected in the matching areas randomly. And then extracts the eigenvectors based on the histogram algorithm. Finally, evaluating the similarity via the histogram of principal curvature direction.The main content of this paper is divided into five parts: The first part summarizes the research background, significance, domestic and international research status and development of the proposed algorithm and the layout of the article; the second part focuses on the theory of curvature and its related estimation algorithms, introducing the concept of integral invariant, Further elaborating the curvature estimation based on integral invariants; in the third part, analyzing the advantages and disadvantages of the traditional mesh model simplification and similarity measurement algorithms, applying the curvature estimation based on integral invariants to the field of simplification and similarity measurement for 3D mesh model; the fourth part is the steps and effects of experiment; the fifth part is the summary of this paper and the prospects for future research work.