A mathematic morphology approach for radial lens correction of document image

Document image captured by hand-held camera suffers from lens distortion with different degrees, and a mathematic morphology based lens correction algorithm is proposed with the text line of document images. First, it uses Niblack adaptive thresholding algorithm to segment the document images, and clusters connected components into text lines with morphology closing algorithm. Then, it uses second order polynomial to fit the central line of text line, and construct the object function of lens distortion. It warps the curved text line to the straight text line, thus, to solve the distortion parameters of document image. Experiments show this method can effectively correct lens distortion of document image with various degrees.