Image analysis – Pattern recognition – Template matching
Reexamination Certificate
1999-05-07
2002-01-29
Au, Amelia M. (Department: 2623)
Image analysis
Pattern recognition
Template matching
C382S311000, C358S452000
Reexamination Certificate
active
06343149
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a character reading system for a document or the like, that verifies or corrects character information read from a document or the like and outputs the correct character information, and more particularly to a character reading system for a document or the like that verifies or corrects, at a central station, data related to the characters of a document or the like sent from a terminal through a communication line and outputs this data as correctly read character information.
2. Description of Related Art
In the past this type of character reading system was disclosed in the publication “OKI DENKI KENKYU KAIHATSU” Vol. 59, No. 4, pp. 23-26, 1992, “OCR Application Systems for Financial Industry Information Systems.”
FIG. 6
is a block diagram schematically illustrating an example of this conventional character reading system.
This system is a character reading system applied to a centralized exchange system for centrally managing exchange. This character reading system
10
comprises reading terminals
14
that read and output information pertaining to the characters of a document
12
or the like, and a central station
20
that centrally processes the information sent from the reading terminals
14
and outputs it as correctly read character information. Usually, the reading terminals
14
are installed at various remote local stations
16
such as “place of business,” and are constituted by an OCR (hereinafter this OCR will be referred to as a remote OCR) or a facsimile device, for example. At the central station
20
, the various information pertaining to characters sent from these reading terminals
14
via a communication line
18
is centrally processed and outputted as correct character information.
With the conventional structural example shown in
FIG. 6
, a remote OCR
14
for reading characters printed or recorded on a document
12
, such as a bank transfer request, is installed at each local station
16
. This remote OCR
14
is a device that reads the image data IMG of the document
12
, converts this into character data DATA, and transmits the image data IMG and the character data DATA to the central station. The remote OCR
14
is connected to a communication network
18
such as an ISDN (Integrated Service Digital Network). Specifically, the character data include so-called “character” data and “numerical” data.
Meanwhile, the central station
20
is furnished with a storage device
22
for temporarily storing information, a correction terminal
24
, a verification terminal
26
, a gateway
28
, a LAN
30
that is connected between these constituent elements
22
,
24
,
26
, and
28
in order to allow information to be passed back and forth between these constituent elements, and a host computer
32
that is connected to the gateway
28
. The correction terminal
24
and the verification terminal
26
are each constituted by a separate microcomputer. The storage device
22
is also controlled by a separate microcomputer.
The communication network
18
is connected to the storage device
22
. This storage device
22
temporarily stores the image data IMG and character data DATA sent from the remote OCR
14
through the communication network
18
in order to correct and verify the character data.
The correction terminal
24
is used to decide whether or not the character data DATA stored in the storage device
22
is correct, and if the character data DATA is incorrect, an operator of this device corrects it. To this end, a conventional correction terminal
24
simultaneously displays on a monitor screen all of the image data IMG of the document
12
along with all of the character data DATA that has been read, including both correct character data and incorrect character data. The operator in charge of correction then makes a visual comparison of the image data with the various character data, and if an error is discovered in the character data, the operator uses the keyboard of the correction terminal
24
to change this incorrect character data to correct character data. The corrected character data DATA is transferred along with the image data IMG to the verification terminal
26
via the LAN
30
.
The verification terminal
26
is a device that verifies that the correction has been properly carried out at the correction terminal
24
. To this end, this verification terminal
26
simultaneously displays on a monitor screen all of the image data IMG of the document
12
along with all of the character data DATA that has been read, including both correct character data and incorrect character data, just as with the correction terminal
24
. The operator in charge of verification then makes a visual comparison of the displayed character data and image data and decides whether the corrected character data is in fact correct. If it is decided that the corrected character data is correct, the operators uses the keyboard of the verification terminal to enter this, and the correct character data is sent to the gateway
28
via the LAN
30
, and is outputted from the gateway
28
to the host computer
32
.
Meanwhile, if it is decided at the verification terminal
26
that the character data is incorrect, this is entered by the operator using the keyboard of the verification terminal, and the image data IMG of the document
12
and the character data DATA are returned to the correction terminal
24
via the LAN
30
.
With a conventional character reading system structured as above, the following operational steps are taken in order to output correct character data. Character data is recognized and read from image data by a single recognition on the basis of a certain character recognition method. Then, regardless of whether the character data that has been read is correct or not, all of the read character data is displayed along with the image data on the monitor screen of the correction terminal
24
, and an operator looks at the character portion of the image data while making a direct visual comparison with the corresponding read character data, and decides whether the character data is correct. Next, after the incorrect character data has been corrected, the correct character data that was not corrected and the corrected character data are transferred along with the image data to the verification terminal
26
, and these sets of data are simultaneously displayed on the monitor screen of the verification terminal
26
. The operator again makes a direct visual comparison of all of the character data displayed on the monitor, and decides whether each set of character data is correct. Only when it has been decided that all of the character data displayed on the monitor screen of the verification terminal
26
is correct, the correct character data is outputted as read character data to the host computer
32
via the gateway
28
.
However, accuracy has been improving for the character recognition itself of the document
12
or the like using OCR or the like. On the other hand, this has not led to a reduction in the correction and verification work that has to be performed by operators at the correction terminal
24
and the verification terminal
26
. The reason for this is that since all of the read character data is displayed on the monitor screen, the operator has to decide the correctness of the character data and perform correction for the character data that needs correcting for all of the character data, regardless of whether correction is needed or not.
SUMMARY OF THE INVENTION
Therefore, an object of the present invention is to provide a character reading system for a document or the like with which the work of correcting and verifying character data read from a document or the like is reduced as much as possible.
The inventors pertaining to this application conducted various research and experimentation in an effort to achieve this object. First, the inventors focused on the fact that there is a variety of character recognition methods. For example, it is known that recog
Au Amelia M.
Dastouri Mehrdad
Oki Electric Industry Co, Ltd.
Rabin & Berdo P.C.
LandOfFree
Document character reading system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Document character reading system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Document character reading system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2845372