A hand-drawn interactive 3D model retrieval method based on high-level semantic attribute understanding

Abstract

The invention provides a hand-painted interactive three-dimensional model retrieval method based on high-level semantic attribute understanding. First, the present invention extracts the global-local semantic attributes of hand-drawn sketches in different styles and categories in a data-driven manner, and defines the semantic attribute space of hand-drawn sketches. Secondly, on the basis of the defined semantic attribute space, the present invention automatically marks the semantic attributes of the three-dimensional models in the database according to their corresponding content features. Finally, the present invention maps the semantic attributes of the hand-drawn sketch and the semantic attributes of the 3D model in the database to the same metric space for comparison by constructing a semantic attribute tree. If the two reach the set similarity, the information of the 3D model is fed back , to complete the search. The method proposed by the invention avoids the deviation caused by the existing hand-drawn interactive three-dimensional model retrieval algorithm that needs to project the three-dimensional model into a two-dimensional view to compare with the hand-drawn sketch.

Description

technical field [0001] The invention relates to a hand-painted interactive three-dimensional model retrieval method based on high-level semantic attribute understanding, which belongs to the technical field of computer graphics and multimedia information retrieval. Background technique [0002] With the rapid development of technologies such as 3D modeling and laser scanning, 3D models have been widely used in many fields such as industrial product construction, 3D video animation, architectural clothing design, and medical information visualization. However, how to quickly and accurately search for the 3D models needed by users from a large-scale 3D model database is still a very difficult task, especially in the context of the current big data era, the types and quantities of 3D models are geometrically The growth of the series, and most of the models are different in function type and appearance structure. The traditional keyword-based and instance-based 3D model retriev...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to address the deficiencies in the existing technology, and propose a three-dimensional model retrieval method based on high-level semantic understanding, so as to solve the problems of low retrieval accuracy due to various hand-painted styles of users, and make the retrieval and query method of hand-painted interaction more convenient. Free expression suitable for users' retrieval needs Based on the above analysis, improve retrieval efficiency and accuracy

Images