Browse Prior Art Database

Document Edge Location and Skew Correction

IP.com Disclosure Number: IPCOM000121636D
Original Publication Date: 1991-Sep-01
Included in the Prior Art Database: 2005-Apr-03
Document File: 4 page(s) / 163K

Publishing Venue

IBM

Related People

Narasimha, MS: AUTHOR [+4]

Abstract

Disclosed is a technique for using image capture data from a high- speed document processor, such as a check reader/sorter, to locate the edge of the document and correct any skew caused by the document being improperly oriented during image capture. Edge location is useful in compressing the data captured so that background data which does not contain document information does not require memory storage space. Skew correction is useful in processing the captured data in order to present a known frame of reference around the data so that it can be more easily determined what the data represents. In some cases, the skew corrected data may be applied to character recognition algorithms to acquire further information from the captured data.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 43% of the total text.

Document Edge Location and Skew Correction

      Disclosed is a technique for using image capture data
from a high- speed document processor, such as a check reader/sorter,
to locate the edge of the document and correct any skew caused by the
document being improperly oriented during image capture.  Edge
location is useful in compressing the data captured so that
background data which does not contain document information does not
require memory storage space.  Skew correction is useful in
processing the captured data in order to present a known frame of
reference around the data so that it can be more easily determined
what the data represents.  In some cases, the skew corrected data may
be applied to character recognition algorithms to acquire further
information from the captured data.

      The captured data is first produced in analog form by a
suitable device, such as a Charged Coupled Device (CCD) or a Time
Delay Integration (TDI) array.  The analog signal is converted to
digital values for each picture element (pel) corresponding to the
data resolution.  Edge detection is accomplished by comparing the
data values with a known background level value to determine when the
data represents document data rather than background data.  The
addresses of the pels corresponding to the document periphery are
stored in memory for future processing use.  Corner location is
determined by noting where two comparatively straight edges
intersect.  In the case of documents with torn corners, as shown in
Fig. 1, the missing document corner is determined from extrapolation
of the existing edge addresses.

      The address locations for the pels along the document periphery
can be used to determine the baseline offsets corresponding to any
chosen section of an edge. This baseline offset characterizes the
skew.  Each of the document edges is divided into multiple sections
of convenient size.  For example, if the data resolution is 200 dpi,
using 100 pels per section allows for a direct relationship between
the skew corrections and the captured pel data.  The skew is
determined by computing the baseline offset in pels/section for each
section along the four edges of the document.  Only the skew values
that are consistent are retained.  This eliminates any inconsistency
due to a document tear or fold.  The retained skew values are
averaged to compute the document skew.

      Correction of the document skew can be accomplished as the
captured data is taken from memory for further processing, such as
thresholding, compression, and resolution reduction.  Fig. 1
illustrates a check 10 which has been scanned by a suitable device to
obtain digital values representing the pels of the image.  The
original scan included areas outside the check and bounded by the
block 11 shown around the check 10. As shown in Fig. 1, the check 10
is not oriented with the intended frame of reference, which is
represented by the boundary lines of the...