3D Model-based Semantic Annotation Management Research And Application

In recent years, along with the development of three-dimensional technology, 3D graphic computing has proven its important position in many major fields. As the 3Drelated resources has been increased recently, 3D scene and object modeling have gained a lot of attractions and supports from various communities. Unfortunately software products that are capable of manipulating these works are still limited in management aspect, making it difficult to extend them to their fullest potential. Users nowadays demand higher requirements from the lower-layer manage components of a 3D software. First, the software products must be able to provide a powerful search engine, which can filter through a large amount of 3D objects to find the desired 3D models or scenes from provided information. Second, the structure, characteristic or type of a 3D object must be analyzed in a reasonable speed to support the traditional semantic search.These requirements lead to a creative idea of annotating the semantic contents to 3D objects. Semantic annotation help users not only maintaining the internal data and structure, but also the semantic contents of a 3D object. These semantic contents include the color, characteristic, type and other information such as the usage for applications. According to the above discussed problems and solutions, this paper design the system can be fully automated semantic annotation process for three-dimensional model. System to meet the user to quickly find the three-dimensional model from the database corresponding semantic annotation information imported. Or system files can be automatically converted into a three-dimensional model of a two-dimensional image. To support semantic information query and search. Or quickly identify the characteristics and classification model for three-dimensional modelTo fulfill the feature of combining all the semantic contents together, the annotating process must meet two required conditions.(1) Proposed an algorithm to extract and analyze the three-dimensional model ofthe visual and semantic information.For the first condition, the method of annotation must have an acceptable accuracy. This means it must have the ability to provide correct and unique information of a 3D object in a specific field. This condition ensures that in the 3D software products, users can easily find their desired 3D objects without filtering too much redundant and unrelated results.(2) Base-on 3D model to implements two auto-semantic annotation methodsThe other condition, which is efficiency-related, becomes necessary when we have to deal with large amount of data. Assume that annotation itself is a complex, timeconsuming and repeated process, the total time which includes duration of user’s manual operation and annotation cannot keep up to the standard requirements of a search engine. High-efficient requirements comprise two things: first, there must be supported tools or computer software to let the annotation process become automatic or semi-automatic, and the semantic contents can have fast inquiry and access; second, the repeated process is able to be recognized to be executed automatically.(3) Base-on 3D model to build a prototype semantic annotation system.According to the above mentioned ideas and algorithms, this paper build a prototype system based on semantic annotation of three-dimensional model of a scene. The system base-on graphics technology to compare the 3D Model with system database of sample models. Which automatically generate semantic annotations for the new model.From the above conditions and requirements, we introduce a Web GL and 3D model-based automatic semantic annotation system. This system is capable of automatically acquire the semantic contents of an object through the data structure analysis or comparing the characteristics between models.