Study On Similarity Evaluation Of 3D Portrait Model For NC Machining And The Application

An important development trend of manufacturing is to develop personalized customized products continually to meet growing market demands. Currently, a kind of personalized customized products with human portraits is gradually forming a potentially emerging industry. This industry mainly contains two parts, the first one is reconstructing three-dimensional (3D) portrait model for that person based on 2D image, the second one is the 3D portrait model was processed with NC machine to obtain 3D personalized customized products with human portraits. A vital point in this industry is the similarity between the product and the modeling object, the key of which is the similarity between the 3D portrait model before NC machining and modeling object. At present, the method of subjective evaluation and expert scoring is used widely, but this method owns some disadvantages, for example, the subjectivity is too strong, and the evaluation results are not stable. Consequently, it's necessary to establish uniform and relatively objective similarity evaluation system. With the support of personalized customized products with human portraits developing project, the similarity evaluation method of 3D portrait model for NC machining is studied. The main study includes the following contents:To begin with, based on the analysis of 3D portrait model's characteristics for NC machining and the need for the similarity evaluation, the similarity evaluation system for 3D portrait model for NC machining is studied, which is composed of two parts, namely, similarity evaluation frame and similarity evaluation process. In addition, the key techniques of similarity evaluation are analyzed.Then, the 3D portrait model registration technology is studied, which is composed by the prepared work before registration and model registration implementation work. Among the prepared work before registration, the acquirement techniques of 3D portrait model, especially the reconstructed 3D portrait model based on images are introduced, and data processing techniques are analyzed in detail. Among the model registration implementation work, in view of 3D portrait reconstructed model and 3D portrait standard model from two different coordinate systems, the model registration technology containing two stages is studied based on the unique structure shape characteristic of face, which are the initial registration based on picking face feature corresponding points and the accurate registration based on the modified ICP Algorithm.And then, on the basis of model registration, a similarity evaluation method based on model error for 3D portrait model is studied. The method is composed by the process of model error analysis and the process of building similarity calculation model. Model error analysis mainly includes two parts, which are model error calculation and model error display. The build of similarity calculation model mainly contains the process of the calculating the model average error and the process of establishing the similarity computing function, and the conditions that similarity functions need to meet are analyzed in detail.Finally, the research results are applied in personalized customized products with human portraits developing project, the similarity evaluation method of 3D portrait model for NC machining proposed in the dissertation is validated the practicality and efficacy.