Video corner-logo detection algorithm based on gradient map of HSV

With the diversification and ever-changing forms of video programs and advertising information, the application of video corner-logo is more and more widely used. Corner-logo detection can be used for video content recognition and interaction, information analysis, monitoring and other aspects of advertisements. An effective corner-logo detection algorithm is the key to corner-logo extraction and recognition and information analysis. In this paper, we analyzed the special problems of corner-logo detection. Corner-logo usually appears shortly and randomly in the video stream, and its time span is uncertain. Corner-logo has various shapes and sizes, appearing in static or dynamic forms. Moreover, there is no priori information for corner-logo detection. These problems make it difficult to find, detect and extract corner-logo. In this paper, we proposed an effective corner-logo detection mechanism, an edge detection algorithm based on HSV gradient map and an optimized weight coefficients allocation method. We detect the scene change frames, and then calculate the three gradient maps of H, S, V components of each frame. We get the combined gradient edge map by weighting and adding the three component maps. By further processing the combined gradient edge maps with time-averaged method we obtained the edge map of corner-logo. Finally, the mathematical morphology algorithm is used to make the boundary connected and the corner-logo region filled. The experimental results show that our mechanism and algorithm can effectively detect and extract the corner-logos in the video streams.