For Full-Text PDF, please login, if you are a member of IEICE,|
or go to Pay Per View on menu list, if you are a nonmember of IEICE.
Context Number Reduction for Entropy Coding of Octree Represented 3-D Objects
IEICE TRANSACTIONS on Information and Systems
Publication Date: 1997/02/25
Print ISSN: 0916-8532
Type of Manuscript: PAPER
Category: Image Processing,Computer Graphics and Pattern Recognition
octree, entropy coding, image coding, 3-D object,
Full Text: PDF>>
The reconstruction of 3-D moving images from transmitted parameters describing position, attitude and shape variation of objects in a virtual 3-D space has been studied as an application of three dimensional (3-D) image communication. The shape information was obtained from a database that was built in advance. Since an appropriate database of 3-D object shapes needs to be developed, efficient storage of the shape data of the actual objects might become a key technology. This paper proposes an efficient entropy coding method of voxel map data obtained with shape measuring equipment. The proposed method converts the voxel map data into an octree and encodes their node information with conditional probability on the state of neighbor nodes sequentially, beginning with the upper hierarchy level. This method has the property of being able to extract information up to a given arbitrary hierarchy level because of its hierarchical characteristic. For implementation, two methods are proposed for reducing the large number of contexts, one uses 3-D isotropism, the other uses sample statistics. The experimental coding results using several sample data sets show that the proposed method can reduce the information volume by about 20% in comparison to the ordinary method using unconditional entropy. The binary voxel map of 512512512 can be represented by approximately 680 kbits.