Browse Prior Art Database

Image Differential Algorithm for Document Reproduction

IP.com Disclosure Number: IPCOM000105268D
Original Publication Date: 1993-Jul-01
Included in the Prior Art Database: 2005-Mar-19
Document File: 4 page(s) / 102K

Publishing Venue

IBM

Related People

Chen, YH: AUTHOR

Abstract

Digital halftoning algorithms used to represent continuous-tone images on bi-level image processing devices (i.e., printers and displays) tend to blur the edges of text and line drawings. However, a thresholding algorithm that produces sharp edges of text and drawings is usually inadequate for continuous-tone images. The described algorithm combines the advantages of digital halftoning for image and thresholding for text by processing text and image portions separately. In addition, the algorithm requires small storage capacity and is fast so that it can be used in small-capacity signal processors in a real- time environment.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Image Differential Algorithm for Document Reproduction

      Digital halftoning algorithms used to represent continuous-tone
images on bi-level image processing devices (i.e., printers and
displays) tend to blur the edges of text and line drawings.  However,
a thresholding algorithm that produces sharp edges of text and
drawings is usually inadequate for continuous-tone images.  The
described algorithm combines the advantages of digital halftoning for
image and thresholding for text by processing text and image portions
separately.  In addition, the algorithm requires small storage
capacity and is fast so that it can be used in small-capacity signal
processors in a real- time environment.

      With this new algorithm, an input document is processed as a
sequence of square blocks.  Each block is determined to be either an
image block or text block.  Text blocks can be processed by a
thresholding algorithm such as fixed or dynamic thresholding, while
image blocks are processed by a digital halftoning algorithm such as
error diffusion or super circle.

      As an example, let high value represent white and low value
black in four rows of input data stored in a buffer.  A 4 x 4 block,
with upper left picture element (pixel) A diagonal to lower right
pixel B, and upper right pixel C diagonal to lower left element D, is
considered to be a text block if:

     a)  Any pixel A, B, C or D> = white cutoff limit; or

     b)  All pixels A, B, C, and D< = black cutoff limit.

     Otherwise, it is considered to be an image block.

     Note:     In b) it is sometimes sufficient to check either A and
B
               or C and D<= black cutoff limit.  Also, with either a)
or
               b), a block of very dark or very bright picture
elements
               is classified as a text block because it is
appropriate
               to process it using the same algorithm as text (e.g.,
               thresholding).

      Input documents are processed in real time.  The processing
flow chart is shown on the Figure.  The system's normal state at
starting block 1 is for text processing.  After initialization 2
(flag is set in OFF position for block 6), data blocks are input to
data buffer 3 and selected for processing at block 4.  Each data
block takes the appropri ate path for processing as text or image at
image/text decision block 5.  Text processing is...