The Description And Retrieval Of 3D Graphics In ScienceML

In order to carry out the transmission, share and retrieval of ScienceWord in Internet, ScienceML was made. ScienceML is a descriptive language of scientific document, and it accords with the specification of XML. 3D graphics are the important elements in ScienceWord, so the descriptive language of 3D graphics is the important part of ScienceML. It is the important precondition of the share of ScienceWord in Internet to implement the retrieval of ScienceWord in Internet. The retrieval of the 3D graphics is an important aspect of the retrieval of ScienceWord.The 3D graphics in ScienceWord, which emphasize the representation of the shape and the effect of the geometrical transform, can be expressed in math equation. According to the characteristics of the 3D graphics in ScienceWord, a tidy, exact and perfect describing way of 3D graphics was introduced. It can recognize the shape exactly, describe all kinds of geometrical transform of the 3D graphics, such as the rotation, zoom and move, and it simplifies the surface attribute which is complex but trivial to 3D graphics in ScienceWord. Then,the methods of creating the XML files describing 3D graphics and the methods of parsing the XML files describing 3D graphics were introduced.On the basis of the descriptive of 3D graphics, the content-based retrieval technology of 3D graphics was discussed. After introducing the principle and the system framework of the 3D graphics retrieval, the primary key techniques of content-based 3D graphics retrieval were analyzed, including the normalization by preprocessing, the feature extraction, the building of index, the similarity measurement and the design of user interface. The system of retrieval of 3D graphics in ScienceML was designed and implemented according to the principle,the system framework,and the methods of the 3D graphics retrieval.The description model of 3D graphics fits with the 3D graphics in ScienceWord.The experiment of retrieval has showed the availability of the retrieval methods which were discussed. The methods can recognize the same graphics and the similar graphics in shape.