Automatic extraction of text regions from document images by multilevel thresholding and k-means clustering | IEEE Conference Publication | IEEE Xplore