A Method of TV Screen Shape Detection Based on Threshold Segmentation

Detecting the boundary of television screen is the basis of content identification for human-computer interaction. Uncertainty exists in TV video image which is captured by mobile phone because of the resolution, color, brightness, background. The reduction of image quality leads to inaccurate identification. We propose a method based on the feature of TV's shape to segment the TV screen indirectly. Our approach aims at extracting the typical feature between the background and objective image. The experimental results demonstrate the effectiveness of the proposed approach for TV screen shape.