Having an integrated platform for implementing and testing previous and new approaches of any scientific problem, is a key for a complete and accurate comparison which itself lead us to the best algorithms. The lack of an appropriate and open source framework for volumetric reconstruction made us to implement an integrated framework for simplifying our development process. This framework can be used for developing new volumetric reconstruction and consistency checking algorithm including shape from silhouette and shape from photo consistency both in dynamic and static scenes. Furthermore reconstructed results can be used as an input for different volumetric presentation approaches which itself is a known topic in computer graphic. There are lots of general reconstruction tasks which are managed by this framework easing the development process. Scene and camera information management, loading images and other required resource, complete control over reconstruction process, presenting reconstructed result and extracting statistical data are among those. Scientists, developers and students can easily develop their ideas in any one of the explained above categories. Then the implementation can be tested and also different statistical data can obtained, giving a remarkable opportunity for comparing different approaches.
Volumetric Reconstruction Host
The volumetric reconstruction - we call it as voxelization - is the process of extracting three dimensional geometry of a scene, using multi different captured views of that scene. The user interface of this framework can load and run all different voxelizer algorithms which are implemented using the framework.
Volumetric Presentation Host
The volumetric presentation - we call it as visualization - is used for presenting the scene, which should be described by voxels, in a graphical manner. There are various methods for presenting a voxel based model including and different implemented visualizer algorithms, using this framework, can be attached to user interface for presenting reconstructed results.
We have used a very open architecture in this framework. Because of this reason any voxelization or visualization can be developed in its own project and library. The framework then binds to the library and takes the control of everything. The late binding process easily is controlled by a configuration file which will be discussed in the next sections.
The framework has been developed using Microsoft Visual Studio® 8 and .NET® framework for Microsoft Windows XP®. Although C# is used for implementation, but all other .NET based languages can be used for more development such as VB.NET, Managed C++.