A method of TV Logo Recognition based on SIFT
Da Pan, Shi Ping · Advances in intelligent systems research/Advances in Intelligent Systems Research · 2013
Abstr act TV logo is an important symbol of TV station.Automatic identification of TV logo can be used in television monitoring and media assets management.By analyzing the characteristics of TV logos, a new algorithm for TV logo recognition is proposed in this paper.First of all, the TV logo information is obtained by applying a special mask image to the captured video frames.Then the coarse matching is performed based on color tone.Further more, the region of TV logo is separated into the master region and sub region and the SIFT keypoints are obtained based on this separation.Finally, the TV logo is recognized by comparing the number of keypoints with a threshold.Experiments results show that the algorithm proposed has a good performance in TV logo recognition, especially for transparent logo.