Restoration of regions occluded by a caption in TV scene

In this paper, we propose a caption detection and removal technique for reuse of TV scenes using multilayer perceptrons (MLP) and genetic algorithms (GA). The technique first detects the captions in a TV scene using a MLP-based caption detector, and then removes the detected captions using a GA-based region remover. In our technique, the caption removal problem is modeled as an optimization problem, which in our case, is solved by a cost function with isophote constraint that is minimized using a GA. The technique creates an optimal connection of all pairs of isophotes disconnected by caption regions. To connect the disconnected isophotes, we estimate the value of the smoothness, given by the best chromosome of the GA, and project this value in the isophote direction. Experimental results show a great possibility for automatic removal of captions in TV advertisement scenes.