A method for segmenting and recognizing text embedded in video and images is proposed in this paper. In the method, multiple segmentation hypotheses of text image are first generated based on a MRF model. Background regions in each hypothesis are then removed by using grayscale consistency constraint (GCC) in a connected component analysis procudure before being processed by an optical character recognition (OCR) software.
Daniel Kressner, Francisco Santos Paredes Quartin de Macedo
Volkan Cevher, Jonathan Mark Scarlett