Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Pruning And Line Approximation of Stroke Data by Using Perpendicular Distance

IP.com Disclosure Number: IPCOM000102531D
Original Publication Date: 1990-Nov-01
Included in the Prior Art Database: 2005-Mar-17
Document File: 3 page(s) / 104K

Publishing Venue

IBM

Related People

Noda, H: AUTHOR

Abstract

Disclosed is a method for extracting the representative points of sampled strokes or continuous vectors, and fitting polygonal lines to them. This reduces the amount of data while preserving the characteristics of the original shapes. Data contained in figures, characters, and images are inputted by tablet, scanner, and video camera, and converted into vector or stroke format. The length of a perpendicular line from each sampled point to the base line, which is connected to the starting and ending points of a stroke, gives the error of line approximation. This length is compared with the threshold to decide whether the bound can be approximated by a straight line or a polygonal line. The threshold guarantees the accuracy of the maximum error of approximation and the similarity to the original shape.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 51% of the total text.

Pruning And Line Approximation of Stroke Data by Using Perpendicular Distance

       Disclosed is a method for extracting the representative
points of sampled strokes or continuous vectors, and fitting
polygonal lines to them. This reduces the amount of data while
preserving the characteristics of the original shapes. Data contained
in figures, characters, and images are inputted by tablet, scanner,
and video camera, and converted into vector or stroke format.  The
length of a perpendicular line from each sampled point to the base
line, which is connected to the starting and ending points of a
stroke, gives the error of line approximation. This length is
compared with the threshold to decide whether the bound can be
approximated by a straight line or a polygonal line. The threshold
guarantees the accuracy of the maximum error of approximation and the
similarity to the original shape.

      Extraction of the features of a stroke can be useful for
compressing the data, and preprocessing can be useful for character
recognition. To fit a long straight line to exact nodes, the system
has to consider the overall shape of a stroke. Conventional methods
for pruning are processed locally. For example, circular masking is
the simplest and most popular method of pruning, but it cannot
extract a long line. Rectangular masking is useful for extracting of
long lines, but is not suitable for finding the exact position of a
node because of the undetectable part of the mask area (Fig. 1). In
order to fit a long straight line to a stroke exactly, it is
essential to extract representative nodes from it. This method uses
the distance from each point to the base line as a threshold for
approximation (Fig. 4). The "accurate mode" uses a fixed threshold,
and guarantees that the maximum error of approximation is less than
the threshold (Fig. 2). However, accuracy prevents extraction of the
representative nodes from a stroke. For example, more nodes are
extracted from a large stroke, even though a large and small stroke
have the same shape (Fig. 3). Usually, the preprocessing part of
character recognition requires the size to be normalized so that it
can extract the features....