Description
Stitching of image content from several sources is an active topic for quite a while, often also referred to as image or photo-mosaics. There are two main research directions which deal with this problem: image stitching and image completion (often also referred to as image in-painting).
This topic addresses students interested in advanced image processing algorithms which aim at one or on both mentioned goals. Recent papers which deal with the problem are for instance (SIGGRAPH 2012):
- Image Melding: Combining Inconsistent Images using Patch-based Synthesis
- Panorama Weaving: Fast and Flexible Seam Processing
This topic focuses on general multi-image compositing where multiple specific theses are possible. Additionally, many aspects of the topics can be adjusted in accordance with the interests of the participant. Students interested in the topic are welcome to contact me at: pm (at) cg.tuwien.ac.at
Possible Topics
- BA: Design and implement a multi-image stitching module in an existing framework.
- DA: Design and implement a GPU/CUDA driven, patch based image completion kernel.
Requirements
- Programming skills in C# and/or C++.
- Knowledge of the real-time graphics pipeline (OpenGL/DirectX)
- Basic knowledge of CUDA/DirectCompute programming is a plus
- Basic knowledge of MATLAB programming is a plus
- Basic knowledge of image processing is a plus
Environment
- Visual Studio 2012
- C#/WPF
- CUDA/DirectCompute/MATLAB
- SharpDX
