Interactive Image Segmentation

energy function representing the optimal segmentation is formulated. The different terms of the function is determined based on the user input (likelihood model) and a prior model. In this approach, the user marks certain pixels as either " background " or " object ". The likelihood is then calculated using this data. Finally, the energy function is minimized using Graph Cuts based technique. The method was implemented and tested using publicly available benchmark data with ground truth ([11,12]). The results and their analysis are also provided in this report.