A graph-based method to remove interferential curve from text image

A general algorithm for eliminating interferential curve in text image was proposed, which converted images into basic graph and super graph first, and then the interferential curve was treated as principal curve to be detected. In the detection, improved DFS algorithm, shortest path algorithm and orientation offset algorithm were used. Finally, the detected curve was removed from the original image, leaving only text in the image. Experiments conducted with a variety of text images showed that this algorithm is effective in removing interferential curve from text image.