A new coding scheme for compressing and manipulating image documents

To detect skew rapidly and to extract connected components from binary docu ment images, a novel approach to image compression is developed. Similar to the CCITT Group 4 compression scheme, this coding scheme exploits the two- dimensional correlation between scan lines and achieves high performance fo r image operations with low complexity. Feasibility of performing connected -component extraction, skew detection, and skew correction in the compresse d domain is presented. Such operations are shown to run faster in the compr essed domain than in the spatial domain.