Title Masking in GUI
Student Fahim Mannan
Mentor Daniel M German
Abstract
While stitching panoramic images, there may be some unwanted foreground object in the final output. Current method of removing those objects, is to use a separate editor and mark the alpha channel of the pixels that we want to remove from the final output. The current process is tedious and time consuming. Therefore, it is desirable to have an application which is fast and easy to use.

The objective of this project is to implement such an editor using concepts developed in interactive image segmentation and texture synthesis works. The workflow will be as follows - the user will first mark the regions that are unwanted with a few brushstrokes and the application will create masks that represents which regions should be kept or excluded in the final image. The user can look at the mask that was created and be able to correct the output interactively.

The core of the implementation will include algorithm that uses Markov Random Field and Graph Cut optimization for determining the optimal mask. The final outcome of the project will be an easy to use editor that takes an image stack along with a set of hard constraints (provided by the user) and finds the best possible mask that satisfies the given constraints.

Further details can be found at - http://wiki.panotools.org/SoC_2008_Masking_in_GUI