PROCEDE ET DISPOSITIF POUR LIRE DES CODES POSTAUX D'ADRESSES SUR DES ENVOIS

METHOD AND APPARATUS FOR READING POST CODES OF ADDRESSES ON MAIL

Abrégé français

L'invention concerne un procédé et un dispositif permettant lire des codes postaux d'adresses d'envois. Selon ce procédé (300), le code postal (300) désigne l'adresse jusqu'à une partie d'une rue importante ou de plusieurs petites rues. Après que des adresses n'aient pas pu être lues de manière univoque dans l'unité de lecture optique (130) lors d'une première tentative de lecture (310), les codes postaux sont d'abord répartis en une partie décrivant sommairement l'adresse et en une autre partie décrivant l'adresse de manière plus précise. Les représentations correspondantes des surfaces d'envois comportant ces adresses sont ensuite transmises à des places de vidéocodage (200) où seules les parties du code postal (300) décrivant sommairement les adresses sont vidéocodées (320) par des opérateurs prévus à cet effet et sont transmises à d'autres places de vidéocodage (200) où les parties restantes du code postal (300) décrivant de manière plus précise les adresses sont vidéocodées (330) de manière asynchrone par des opérateurs prévus à cet effet. Pour finir, les parties de code postal appartenant à un envoi déterminé, qui sont stockées temporairement et vidéocodées sont regroupées (340) en un code postal vidéocodé complet.

Abrégé anglais

The invention relates to a method and device for reading postal codes of
addresses on postal articles, in which the postal code (300) designates the
address up to the portion of a major street or of a number of minor streets.
When addresses could not be clearly read during a first reading attempt (310)
in the OCR unit (130), the postal code is first of all divided into a part
that roughly depicts the address and into a part that depicts the address more
precisely. The corresponding liknesses of the postal article surfaces to these
addresses are then transmitted to the video coding stations (200) where only
the parts of the postal codes (300) that roughly depict the addresses are
video-coded (320) by operators provided therefor and are transmitted to
additional video coding stations (200) where the remaining postal codes (300)
that depict the addresses more precisely are asynchronously video-coded (330)
by operators provided therefor. Afterwards, the buffered, video-coded postal
code parts, which respectively pertain to a particular postal article, are
combined (340) to form a complete, video-coded postal code.

Revendications

-9-
Claims
1. A method for reading post codes of addresses on
mail, in which the images of the mail surfaces which
have the addresses are recorded, stored and supplied to
an OCR unit (130) for identification of the addresses
and generation of appropriate address codes, in which,
if OCR unit (130) reading results are not unique, the
relevant images are video-coded at at least one video
coding location (200), and in which the post code (300)
identifies the address down to a part of a relatively
long road or to a number of small roads, in which
- if it has not been possible to uniquely read
addresses in the OCR unit (130) in a first reading
attempt (310), the post codes are each first of all
subdivided into a part which describes the address
roughly and a part which describes the address in more
detail and the corresponding images of the mail
surfaces with these addresses are transferred to video
coding locations (200) where only those parts of the
post codes (300) which describe the addresses roughly
are video-coded (320) by operators who are provided for
this purpose,
- and the images of the mail surfaces with the
corresponding addresses are transferred to further
video coding locations (200), where the rest of the
post codes (300) which describe the addresses in more
detail are video-coded (330) asynchronously with
respect to them by operators who are provided for this
purpose, and the video-coded post code parts which have
been buffer-stored and are each associated with a
specific mail item are then joined together (340) to
form a complete, video-coded post code.
2. The method as claimed in claim 1, with that part
of the post code (300) which describes the address
roughly indicating the destination area.

-10-
3. The method as claimed in claim 1, with that part
of the post code (300) which describes the address in
more detail identifying the road, a part of the road in
the case of long roads, or a number of small roads.
4. The method as claimed in claim 1, with the video-
coded post code elements being associated with the
respective mail items or images on the basis of
identification numbers, and being joined together to
form complete post codes (300).
5. The method as claimed in claim 4, with the
identification numbers being applied to the mail items
and being read together with the addresses.
6. The method as claimed in claim 4, with the
recorded images of the mail surfaces being stored under
identification numbers.
7. The method as claimed in claim 5, with the
identification numbers being applied as machine-legible
codes to the mail items.
8. The method as claimed in claim 7, with the
identification numbers being applied as a bar code to
the mail items.
9. The method as claimed in claim 1, with a preview
coding method being used for the video codings.
10. The method as claimed in claim 1, with the
complete, video-coded post code being supplied to the
OCR unit (130) for a second reading attempt (350),
including the video-coding results, in order to
validate the result.
11. An apparatus for reading post codes of addresses

-10a-
on mail items using an automatic address reading
system, which
- has an apparatus for obtaining images of mail items
(120),

-11-
- has an OCR unit (130) for automatic evaluation of the
images which have the address information on the mail
surfaces,
- has video coding locations (200) for video-coding of
the images which contain the address information on the
mail surfaces,
- has an image controller (170) for controlling the
data flow between the OCR unit (130) and the video
coding locations, in which the image controller (17) is
designed such that
- if it has not been possible to uniquely read
addresses in the OCR unit (130) in a first reading
attempt (310), the post codes are each first of all
subdivided into a part which describes the address
roughly and a part which describes the address in more
detail and the corresponding images of the mail
surfaces with these addresses are transferred to video
coding locations (200) where only those parts of the
post codes (300) which describe the addresses roughly
are video-coded (320) by operators who are provided for
this purpose,
- and the images of the mail surfaces with the
corresponding addresses are transferred to further
video coding locations (200), where the rest of the
post codes (300) which describe the addresses in more
detail are video-coded (330) asynchronously with
respect to them by operators who are provided for this
purpose, and the video-coded post code parts which have
been buffer-stored and are each associated with a
specific mail item are then joined together (340) to
form a complete, video-coded post code.
12. An apparatus as claimed in claim 11, with the
image controller (170) being designed such that the
respective assembled video-coded post code (300) is
supplied to the OCR unit (130), for a second reading
attempt, including the video-coded post code (300) for
validation of the result.

Description

CA 02432610 2003-06-20
GR 2000P23858
s
Description
Method and apparatus for reading post codes of
addresses on mail
The invention relates to a method and an apparatus as
claimed in the precharacterizing clauses of the
independent patent claims.
Systems for automatically reading addresses (OCR) are
well known in the field of letter processing and are
described, for example, in DE 195 31 392. Modern OCR
letter sorting systems can achieve processing rates of
10 letters per second, that is to say 36 000 letters
per hour, or more. However, the identification
reliability varies sharply with the type of writing and
the overall quality of the address information applied
to the surface of the letters. If identified
successfully, the relevant letter can be provided with
a machine-legible bar code. This bar code allows
further machine processing down to any desired sorting
level. In particular, the use of bar codes allows the
letters to be sorted down to the individual post round
sorting level, in which letters are sorted on the basis
of the sequence of their distribution by the postman.
Since the identification rates of automatic reading
systems vary widely, they must be assisted by various
forms of manual intervention.
In this case, the entered address is coded and is
converted by means of a directory to a machine-legible
sorting bar code, which is applied to the mail item.
In this case video images of the mail items are
provided to operators for coding, in video coding
systems. This may be done online and offline. In the

CA 02432610 2003-06-20
GR 2000P23858 - la -
case of online video coding systems (OVS), the video
image is shown to the operator, while the physical

CA 02432610 2003-06-20
GR 2000P23858 - 2 -
mail item is held in delay sections. In these delay
sections, the mail item is normally stopped in its
movement for a time period which is sufficient for the
OVS operator to enter the necessary sorting information
for the relevant image. The normal delay sections allow
a delay of between 10 and 30 seconds. The longer the
delay section, the greater are the costs and the
requirements for maintenance, and the physical size of
the system.
When using OVS, the available time is frequently
sufficient only for one careful input of the post code
( PK) .
In the case of offline coding systems, such as those
described in US Patent Specification 49 92 649, mail
items with unidentified addresses are provided with
additional information, a tracking identification
(TID). The unidentified mail items are stored
externally, while the images of these mail items are
presented to operators for coding, without there being
any time restrictions. The mail items are then passed
over TID readers. The TID is linked to the entered
address information. Based on this, conventional bar
code sorting information can also be applied to the
mail item, so that the relevant mail item can be
processed in the same way as normal OCR-read mail
items. Although the offline video coding method
represents an effective method for coding all the
components of the address, it requires additional
capacity for the further processing of mail items which
are provided with addresses that have not been read,
and demands appropriately complex logistics.
In the case of post codes with six or more characters
which describe the address down to the road or to a
number of small roads, the input for the video coding
process is made using the left and right hands.

CA 02432610 2003-06-20
GR 2000P23858 - 3 -
The invention as specified in claims 1 and 11 is based
on the object of shortening the video coding process
for reading a post code, which is not identified
uniquely by an OCR unit, in a mail address which
identifies the address at least down to the road.
This is achieved by subdividing the post code into a
part which describes the address roughly and a
remaining part which describes the address in more
detail in each case. Each part is video-coded by
operators provided for this part. After one failed
attempt by the OCR unit to read the respective address,
the left-hand part of the post code, which describes
the address roughly, is video-coded at an appropriate
video coding location and, asynchronously thereto, the
remaining right-hand part of the post code, which
describes the address in more detail, is video-coded at
an appropriate video coding location.
After this, the buffer-stored and video-coded post code
parts which are associated with the respective mail
item or image are joined together to form a complete,
video-coded post code.
The time for video coding of each part of the post code
is in this case shorter than if the post code had been
video-coded in total by one operator, since each
operator has to enter only some of the symbols.
Advantageous refinements of the invention are described
in the dependent claims.
It is advantageous if that part of the post code which
describes the address roughly identifies the
destination area, and that part of the post code which
describes the address in more detail identifies the
road, a part of the road in the case of long roads, or
a number of small roads.

CA 02432610 2003-06-20
~ GR 2000P23858 - 4 -
In a further advantageous embodiment, the video-coded
post code elements can be associated with the
respective mail items or with their images on the basis
of identification numbers.
For this purpose, it is advantageous to apply the
identification numbers to the mail items and to read
them together with the addresses, in which case it is
particularly advantageous to apply the identification
numbers such that they are machine-legible, in
particular in the form of a bar code.
It is also advantageous to store the recorded images of
the mail surfaces under identification numbers.
In order to increase the coding rate, it is
advantageous to use a preview coding method in which
images of a number of mail items that follow one
another are displayed on the screen at the same time.
In order to check the result of the video coding
process, it is advantageous to supply the entire coding
result from the OCR unit to a second reading attempt,
including the coding result. The confirmation is then
produced on the basis of the address data bank, in this
case.
The invention will be explained in more detail in the
following text with reference to the drawings of one
exemplary embodiment, in which:
Figure 1 shows a schematic illustration of an
apparatus for reading post codes,
Figure 2 shows the method sequence on the basis
of a flowchart.

CA 02432610 2003-06-20
GR 2000P23858 - 4a -
Figure 1 shows a schematic illustration of a letter
distribution system by means of which the method
according to the invention can be carried out . The OCR
letter sorter 100 comprises a feed

CA 02432610 2003-06-20
GR 2000P23858 - 5-
device 110, which draws successive mail items from a
magazine and transports them at a rate of about 10 mail
items per second to a high-resolution video scanner
120. The mail items are then transported in a delay
section 121. The mail items normally have address
information on their surface. The address information
from the images of the mail items, as obtained by the
video scanner 120, is then evaluated in an OCR unit
130. If complete evaluation is carried out, a bar code
printer 150 is actuated, and the mail item is provided
with an appropriate bar code for subsequent sorting
into sorting compartments 160. The OCR unit 130
comprises one or more microprocessors 131 with an
associated memory 132 , in order to store images of the
mail items . The OCR unit also contains an address data
bank 134 with post codes, town names and road names,
and possibly with other address-related information.
During the evaluation of images which have address
information, a feature-controlled reduction of the
entry obtained from the address databank 134 is carried
out, so that a sort of dictionary element is produced.
Credibility levels are in this case allocated in
individual entries, so that a number of data items of
correctly identified addresses are produced during the
evaluation process. The apparatus also contains an
image controller 170 as well as a number of video
coding locations 200, which are connected to the image
controller 170 directly or by means of a local area
network (LAN) 171. If the OCR evaluation of an image
was not, that is to say, completely successful, this
image is transferred from the OCR unit 130 to the image
controller 170, which firstly controls the TID bar code
printer 151 and secondly sends the appropriate image to
one of the video coding locations 200. The TID bar code
printer 151 applies an identification code TID to the
corresponding mail item, which identification code TID
makes it possible to subsequently link the evaluated

CA 02432610 2003-06-20
GR 2000P23783 - 5a-
address information with the physical mail item. The
images are in this case preferably evaluated offline,
although, in principle,

CA 02432610 2003-06-20
GR 2000P23783 - 6 -
online evaluation by means of video coding is also
possible, provided the delay time is sufficiently long.
In the latter case, the TID may also be applied to the
mail items at a later time, that is to say when the
video coding has not led to complete evaluation within
a specific, predetermined time.
In order to carry out the method as described in more
detail in the following text, the image controller 170
is designed such that the images are transmitted with
address information which has not been read uniquely by
the OCR unit 130 to in each case two video coding
locations, at each of which only a part of the post
code is video-coded, and such that the coded post code
parts of the relevant image are combined once again,
after buffer-storage, to form a complete post code.
The recorded and stored image of the respective mail
item with the address, containing a post code 300 which
identifies the destination area and the road, the part
of a road in the case of long roads or a number of
small roads, is supplied to the OCR unit 130 for a
first automatic reading attempt. This post code 300 is
in this example a six-character code in canonic form
with a sequential character sequence ANA NAN, where A
is a letter (alphabetic character) and N is a number
(numerical character).
The specific embodiment of the post code is, of course,
not restricted to this example.
If the address can be read uniquely in the OCR unit 130
in the first reading attempt 310, the identification
task is complete. If not, the image together with the
post code 300 is transferred to a video coding location
200 for video coding of the left-hand ANA part, which
characterizes the destination area, at which video

CA 02432610 2003-06-20
GR 2000P23783 - 6a -
coding location 200 an operator for the left-hand post
code parts video-codes 320 the left-hand ANA part. To
do this, the left-hand part of the post code can be
especially identified on the screen at the video coding
location 200 for example by a colored background

CA 02432610 2003-06-20
GR 2000P23783 - 7 -
or by a boundary, as shown in Figure 2. The coding
result is then transferred to the image controller 170.
Furthermore, the image is transmitted with the post
code, for video coding of the right-hand NAN part which
characterizes the roads, to a further video coding
location 200, at which a responsible operator video-
codes 330 the right-hand NAN part asynchronously with
respect to the left-hand NAN part. In this case as
well, that part of the post code which is to be coded
is advantageously identified on the screen. The result
of the video coding is then likewise transmitted to the
image controller 170. In the image controller 170, the
coding results are buffer-stored under an
identification number for the respective image, and are
combined 340 to form a complete video-coded post code,
with the image controller 170 being informed during the
transmission of the coding results as to what part
(rough or more detailed descriptive) this is, so that
the combination can be carried out in the correct
position. The complete video-coded post code is then
transmitted to the OCR unit 130 for a third reading
attempt 350. Since this reading attempt 350 is now
carried out with the entire entered post code, the
object of the OCR unit 130 in this case is to compare
the post code with the content of the address databank,
and possibly with the other address parts that have
been read, and to validate whether the post code 300
has been entered correctly.
By splitting the post code 300 into two parts, which
are video-coded separately, and by the linking with the
OCR reading process, this results in the following
advantages:
1. The shorter sequence of characters to be coded for
an operator (video coding location) results in a
very much lower error rate and makes the preview

CA 02432610 2003-06-20
GR 2000P23783 - 7a -
coding process simpler, during which the
address images of a number of mail items
following one another are displayed at the same
time.

CA 02432610 2003-06-20
GR 2000P23783 - 8 -
2. The training time required for entering the
characters in the preview mode is greatly reduced.