Scanned Document Images Skew Correction Based on Shearlet Transform

During the documents scanning process, a few degrees of skew is unavoidable. Which will cause some adverse effects on the documents identification, such as Optical Character Recognition (ORC). This paper presents a scanned document images skew estimation and correction algorithm based on Shearlet transform. Shearlet transform offers very good time-frequency localization and direction selectivity. It is possible to detect the skew orientation of the document images accurately. Experimental results show that the proposed algorithm has a high accuracy rate of skew estimation even the scanned document images contain noise or include some pictures or diagrams.