Unwarping scanned image of Japanese/English documents

We present methods for eliminating or reducing the distortion in a scanned image. Aspects of the present paper allow for the automatic pruning, de-skewing, and unwarping of an image using boundary document layout information. Here, two dominant top/down baselines are selected, in part, by examining the letter spatial locations on boundary baselines rather than examining the entire document layout. It shall be noted that present method is robust enough to handle many types of content, including different languages: Japanese and English, as well as documents with different layouts. The algorithm is applied to images obtained from bound documents and flat documents.