Today, one of the key areas of research in the field of telecommunications is the task of compressing multidimensional data sets to ensure their efficient transmission over communication channels that have limited bandwidth.
To solve this problem, it is necessary to develop highly effective methods and algorithms for reducing the redundancy of multidimensional data arrays transmitted over communication channels.
The dissertation work is devoted to the decision of an important and actual question in the field of modern telecommunication systems - to development of a technique, mathematical models and effective technologies of formation and reduction of redundancy of the three-dimensional images which are transferred on communication channels.
In the dissertation the important scientific and practical problem in is solved branches of modeling - The discrete matrix model of representation of spatial objects for problems of synthesis, reconstruction and the analysis of their form is offered. The model is based on discrete data about the object which, for example, are raster images or readings of spatial scanners. In contrast to such voxel models, matrix models do not describe the volume but the surface of objects, and while maintaining the advantages of voxel models, such as simplicity and regularity of structure, eliminate their inherent redundancy. It is shown that by storing information about the form sufficient for object visualization, the matrix model can occupy 4-8 times less memory compared to voxel (the comparison was made for models in the VOX format of the MagicaVoxel package). The conditions under which the matrix model remains more economical than the voxel model is established, and it is shown that these conditions are fulfilled for practically significant cases.
In the process of performing the work for the first time the following scientific results were obtained.
In the dissertation work the technique of accessible and inexpensive technologies of formation of three-dimensional images of multistory buildings in which for definition of height of the building is offered, ultra-high spatial resolution satellite images are used.
In the dissertation work the important scientific and practical problem in the field of reduction of redundancy of three-dimensional images was solved - the discrete matrix model of representation of spatial objects which allows to reduce the size of the files transferred on communication channels is offered.
Also, the result of the presented work is a matrix model, which, as well as existing voxel models, based on the discreteness of the original data about the object and their simplicity, in contrast to the latter, takes into account only the surface shape of the spatial body and thus eliminates redundancy. associated with the need to store information about the volume of the object.
Comparison of the proposed matrix model to represent those spatial objects as voxel models in VOX format used in the MagicaVoxel package, confirms their cost-effectiveness Depending on the specific form, DM requires a memory volume that is 4-8 times less than for VM.
The limitations of the proposed model include the lack of accounting for transparency, which is overcome in voxel models by using voxels of different optical densities. This issue needs to be addressed in further research.
As a result, one of the areas of future research is to find an effective transformation of the matrix model into a description of the form based on graphical primitives of modern modeling systems (by the way, Blender) or visualization libraries (by the way, OpenGL).
In addition, the studies require the prospects of using the proposed DMM in applications that use the analysis of the surface of objects for their reconstruction and identification.
