US Rapporteur Report

linksnewsAI and Robotics

Oct 18, 2013 (3 years and 10 months ago)

140 views


1

IPC Revision WG


Definition Project


Project: D167

S
ubclass
: G06K

US Rapporteur
Report

Date

:
Octo
ber
8
, 2
00
9


Ref:

Annex
2
9
,
2
7

April

200
9
:
E
P

Comments


Annex
30
,
25

May

200
9
:
JP Comments


Annex
31
,
29

May
20
0
9
:
SE

Comments


Rapporteur

wants to thank
E
P
,
JP
,

and
SE

for their
c
omments, and apologizes for the
late report.



1
/ Referring to
E
P

C
omments


A/
EP
would like the phrase “designing computerised pattern recognition systems” to be
reinserted in the defin
ition statement of G06K 9/00

on the grounds that “EPO cannot
see why this would not relate to pattern recognition nor can see what field would have
precedence and cover better the design of pattern recognition systems tha
n

the field of
pattern recognition
itself”
.
Upon further review,
Rapporteur still believes that this subject
matter should not be covered in G06K 9/00

since it would appear to expand the breadth
of the G06K 9/00 title as presently written (Methods or arrangements for reading or
reco
g
nising

printed or written characters or for recognizing patterns, e.g. fingerprints)
.
Unless there is an in depth analysis of the documents to warrant the subject matter
being added to the definition, Rapporteur would prefer not include it at this time. Furthe
r
document analysis
could also be done in
a Maintenance project

once this Definition
project is completed
.


B/ EP states that “Computer Vision has a very strong overlap with image recognition so
that it is normally associated with image recognition and tha
t it would not make much
sense to separate it from pattern recognition”.
In view of the many subgroups under
G06K9, EP feels the need to make the distinction with G06T clearer. EP therefore
proposes to
add “computer vision” in the glossary as follows:


c
omputer/machine
vision

computations involved in the recognition of objects or
of scenes or in the understanding of scenes when these
computations are applied to data obtained from one or
more image cameras, i.e. when the images of the objects
or scene are
optically acquired without physical contact
the sensor(s) and the objects


While Rapporteur agrees with EP that computer vision has a very strong overlap with
image recognition,
computer vision also has a strong overlap with other sub domains
such as scen
e reconstruction, event detection, tracking, learning, indexing, motion
estimation, and image restoration. With this in mind, Rapporteur believes
EP’s
definition of computer/machine vision
(
computations involved in the recognition of
objects or of scenes

…)
is more limited/specific tha
n

the actual definition
.
Since the

2

term
computer/machine vision

is not exclusively used in the art covered by G006K 9/00
as indicated above, Rapporteur will not present it as a glossy term.


C/ EP proposes
to include “data f
usion” in the G06K 9/00 definition statement, since
G06K 9/62 is the most appropriate place (under item 92 of the IPC Guide).
Item 92 of
the IPC Guide states “… that the technical subject of an invention may be expressed as
different categories of subject

matter. If for one of these categories, no distinct place
has been identified by the titles of the Classification for a particular technical subject
matter, the most appropriate place existing for the other categories is used for
classifying (see paragra
phs 93 to 99, below, for specific situations)

.
I
tem 96 of the IPC
Guide

presents a specific situation
,
wherein if
a place directed to a process for making
or treatment of products does not exist, the
n the

making or treatment of products is
classified in
the place for the apparatus performing the process.

Therefore, it is clear
from item 92 of the IPC Guide that the technical subject of an invention may be
expressed as a different category when the one category most relevant to the technical
subject of an

invention does not exist. Such is not the case with EP’s proposal
of
“data
fusion”

in G06K 9/00
.
While Rapporteur agrees with EP that pattern recognition
strongly involves data fusion, it is however that
data fusion, per se, should not be
covered under
G06K 9/00.
Data fusion is not limited for use in pattern recognition
systems, as the general concept of data fusion involves the combining of data
, images,
or signals

from multiple sources
.

For example, the subject matter of data fusion for use
in a part
icular art could be classified in IPC D05B 19/00, G06F 7/32, or H04M 11/06.
Rapporteur therefore does not approve EP’s proposal to indicate “data fusion” in the
G06K definition statement.


D/ EP proposes to amend the definition statement of G06K9 as follo
ws:




Fundamental,

computerised recognition functions
operating on general data,
general images or general signals (e.g. represented by n
-
tuples such as
vectors):

statistical
feature

extraction
,
source separation,
clustering,
classif
ication, pattern matchi
ng (G06K9/62, 70
-
76).



Data fusion, i.e. computations for merging heterogeneous data, e.g. various
kinds of images, signal, or data derived therefrom (features, classification
results).



Computational functions operating on images, may the images be acquired

by
contact sensors, or may the images be acquired by cameras as in the case of
computer vision, which functions could be part (modules, processing steps or
stages) of an image recognition system, even when the function is taken alone
without the global re
cognition context (G06K9/20
-
60, 64,68):

-

whole artificial vision systems, e.g. systems modelling (mimicking human or
animal) visual attention (G06K9/60, G06K9/78
-
82);

-

computations for image acquisition (G06K9/20) such as those of low
-

or mid
-
level compu
ter vision: deriving three
-
dimensional information from images of
scenes subject to specific illumination patterns; aligning or centering the
image
-
pickup (G06K9/32), e.g. computations for camera calibration such as
those related to the projection matrices
; aligning images (G06K9/32), e.g.

3

computations for image registration, for stitching neighbouring images, for
aligning multiple image views based on projective geometry or for stereo
matching or epipolar geometry; image centering (G06K9/32), e.g. object
d
etection or tracking in images; image segmentation (G06K9/34); detecting
edges or contours from the image (G06K9/46);

-

certain image processing functions such as statistical computations for image
noise filtering (G06K9/40);

-

computations, such as those
of mid
-

or high
-
level vision, for extracting image
features, e.g. descriptors (G06K9/46) such as those based on object contour,
e.g. deriving shape models (G06K9/48); deriving mathematical or geometrical
properties of the image (G06K9/52);

-

computations,

such as those of high
-
level vision, for image recognition, e.g.
image matching (G06K9/64, G06K9/68).



The above
-
mentioned computations when specific for specific kinds of images
or movements:

-

printed or written characters, images of documents;

-

biometri
c patterns except speech: fingerprints; images of the whole (human
or animal) body or of parts thereof such as faces, eyes, vascular patterns;
human movements such as handwritten signatures or gait;

-

scenes and audiovisual content: aerial images, indoor i
mages, outdoor
images, surveillance images, road images, television content;

-

industrial images;

-

medical images, microscopic objects.



Specific acquisition arrangements for certain specific patterns such as: printed
characters, handwriting, fingerprints;

also faces, body or eyes but when
specific for computerised recognition.



Means (e.g. specially adapted user interface) for correcting errors produced by
computerised pattern recognition
(G06K9/03)
or for designing computerised
pattern recognition systems

(G06K9/62)
; means for assisting humans in
presenting patterns (e.g. positioning a finge
r) for computerised recognition
(G06K9/20).


Rapporteur believes the
subject matter as presented in the first, fourth, and fifth bullets
above
are already covered in the

current G06K 9/00 definition statement
. The subject
matter as presented in the second
and sixth
bullet
s

could not be accepted for reasons
as explained in the above
item
s

C

and A, respectively
.
Rapporteur believes the subject
matter as presented in the t
hird bullet above can not be accepted since it contains
technologies (systems modeling, deriving three
-
dimensional information from images of
scenes subject to specific illumination patterns, computations for camera calibration
such as those related to the

projection matrices
)
beyond the scope of G06K 9/00.







4

In summary, Rapporteur would prefer not to go beyond the current
G06K 9/00 definition
statement

without further analysis. The G06K 9/00 definition statement has been
maintained, and

is reproduced
below

as follows
:




The (computerised) recognition of specific patterns, e.g. printed or written
characters, documents, scenery, audiovisual content, medical or industrial
images, biometric patterns (e.g. gait).



General (computerised) recognition functions
such as extracting features,
clustering, classifying or matching patterns, segmenting images, or separating
(signal or image) sources.



Image acquisition for specific patterns, e.g. fingerprint or iris sensors.



Image pre
-
processing, e.g. image thinning, noi
se filtering.



Means (e.g. specially adapted user interface) for correcting errors produced by
computerised pattern recognition; means for assisting humans in presenting
patterns (e.g. positioning a finger) for computerised recognition.


2
/ Referring to
JP

comments

A/
JP

[item 1 of Annex 30]
presents arguments that “the subject matters to be covered
by the description of “or for recognising patterns, e.g. fingerprints” in the current title of
main group G06K9 are running over the scope of subclass G06K as w
ell as of main
group G06K 9/00”

and “the term “data” in the phrase of “recognition of data” in the title
of subclass G06K means such data as printed or written characters that were artificially
generated on the assumption that the recognition will be made
thereof. It does not cover
fingerprints or natural images as in computer vision that were not artificially generated”.
JP further contends that “in compliance with the general rule of IPC that the scope of
main groups must be limited within the scope of s
ubclass, “Methods or arrangements
for reading or recognising printed or written characters” should be the substantial scope
of G06K 9/00” and

For some subject matters for which the field of application is not
determinable, classification should be given i
n G06T as being related to the image data
processing in general”.


JP therefore would like to propose the following definition statement for G06K 9/00:

-

General recognition functions such as extracting features, clustering, classifying
or ma
t
ching pattern
s or separating image sources for recognizing printed or
written characters.

-

Specific acquisition for reading and recognising printed or written characters.

-

Specific pre
-
processing for recognising printed or written characters.


Rapporteur, in item 1 of An
nex 20, ha
d

addressed JP
’s proposal. However, Rapporteur
has
revisited JP’s
proposal and
concerns, but
it is still that
Rapporteur can not adopt
JP’s proposal
for the same reasons as stated in item 1 of Annex 20. To reiterate, an
important aspect of “rec
ognition of patterns” will be missed if JP’s definition statement

5

proposal is implemented, as the G06K 9/00 title “Methods or arrangements for reading
or recognizing printed or written characters or for recognizing patterns, e.g. fingerprints”
clearly incl
udes such subject matter.

Regarding JP’s contention that the classification of subject matter relating to the
recognition of natural images could be given in G06T, Rapporteur believes it would be
improper to classify such subject matter in G06T.
G06K 9/
64 and G06K 9/68 clearly
covers image recognition

features
. Further, a
s expressed by
EP
[Annex
18]:

G06T covers image data processing
in general
,
whereas

pattern recognition, regarded
as a special
-
adaptation of the image
data
processing in general, is ex
cluded by the
existing references after G06T title and more explicitly by the Note (2) after G06T:

Note (2) This subclass [G06T]
does not cover
:

-

reading or recognising printed or written characters or recognising patterns, e.g.
fingerprints, which is cov
ered by subclass G06K

Therefore, it
i
s clear that
image and
pattern recogn
i
tion
features
should be
covered in
G06K 9/00.


B/
JP [item 2 of Annex 30] would like the first item of Defin
i
tion Statement for subclass
G06K be rectified so as to avoid containing
other subject matters than those intended
by “Recognition of data”.

Rapporteur does not believe that there is any indefiniteness
pertaining to the first item of Definition Statement for subclass G06K and therefore no
change is
deemed
required.


3/ Referri
ng to SE Comments


A/ SE believes the border between G06K and G11 are not clearly defined. SE remarks
that there is a reference in G11B to G06K “recording or playback apparatus using
mechanically marked”, but G06K 19/00 relates to digital markings. Rappo
rteur believes
this point brought up by SE requires some further attention or review to the G11B
Definition, and which is not in the realm of the current
G06K
definition project. Perhaps
this could be better handle
d

in
a Maintenance project, Proposals for

improving the IPC,
e.g. M013, or re
-
activating Definition project D052 for G11B
.


B/
Regarding
G06K
(
informative references
)
,
SE questions the need to possibly include
an additional referen
ce

to places which may be of interest for searching regarding ac
tive
transponders in G06K

and G06K 19/00
. Rapporteur believes
the informative reference
to H04B 1/59 “Responders; Transponders” as currently provided in the G06K definition
covers active transponders.

Rapporteur has
however
added the informative referenc
e
to H04B 1/59 in G06K 19/00, as suggested by SE.



6

C/

Regarding
G06K
(
special rules of classification
)
, Rapporteur has moved the last rule
“G06K 9/58 takes precedence over groups G06K 9/38


G06K 9/54”
under the
special
rules of classification within this
subclass

section of subclass G06K to the

main group
G06K 9/00
,

as suggested by SE.


D/ Regarding
G06K (glossary of terms)
, SE believes “data” should only appear below
the
glossary of terms
, and not below the
synonyms and keywords

section of the
definition
s

since terms found in the scheme or definition statement should appear below
glossary of terms
. Therefore, Rapporteur has deleted the explanation of “data” in the
synonyms and keywords

section of the definitions as suggested by SE.


E/
Regarding
G06K 1/00

(definition statement)
, SE believes the phrase “the like” should
be changed to “similar record carriers”
.

Rapporteur believes the phrase in question is
commonly used in the IPC and does not raise any issues of indefiniteness (see item 65
of the IPC Guide)
. Rapporteur therefore would prefer to maintain the definition
statement with the phrase “the like”.

SE makes comments that “specific” should be
change
d

to “specially adapted”.
Rapporteur believes
such change is in compliance with item 6
4

of the IPC Guid
e, and
therefore has adopted the change as proposed by SE.


F/ Regarding
G06K 1/00 (references relevant to classification in this group)
,

SE
indicates that the reference to G06K 7/00 is merely an informative reference since
sensing cannot be classified at

G06K 1/00.
Rapporteur however believes there is an
overlap of subject matter in both areas,
since
G06K 1/04 covers “controlled by sensing
markings on the record carrier being punched”

and G06K 7/00 is directed to “methods
or arrangements for sensing reco
rd carriers”
.
Rapporteur
therefore will
maintain the
limiting reference to G06K 7/00

in G06K 1/00.



G/ Regarding
G06K 3/00 (definition statement)
, Rapporteur does not believe the
definition statement is in accurate as purported by SE since it was derived
from the
title
of G06K 3/02.


H/ Regarding
G06K 5/00 (definition statement
)
, Rapporteur does not believe the
definition statement is inaccurate as purported by SE since it was derived from the titles
of G06K 5/02 and G06K 5/04.


I/
Regarding
G06K 7/00 (def
inition statement)
, SE
suggests adding “e.g.” before the list
to fully cover the main group and that it is necessary to provide a definition for “sensing”

as “reading the content of the information carrier”
. Rapporteur
has adopted the
add
ition
of


,
e.g.”

before the list

as suggested by SE
. Rapporteur
however
believes
it would be

7

more accurate to define
“sensing”
as “reading the content of the record carrier”
, which


s
uch change has been adopted in the
glossary of terms
.




J/
Regarding
G06K 7/00 (inf
ormative references)
, SE is correct that “Transponders
powered by received radio waves, e.g. passive transponders” should refer to G01S
13/75.
Rapporteur thanks SE for identifying the typo and correction

has been made
(see Annex 33).


K/
Regarding
G06K 7/
00 (references relevant to classification in this group)
, SE is of the
opinion that the reference to G06K 1/00 is merely an informative reference since
marking cannot be classified at G06K 7/00. Rapporteur however believes
the reference
is proper as there

are overlapping subject matter
between
G06K 1/00 and G06K 7/00.
For example,
G06K 1/04 covers “controlled by sensing markings on the record carrier
being punched” and G06K 7/06 covers
“by means which conduct current when a mark
is sensed or absent, e.g.
contact brush for a conductive mark”. Rapporteur will
therefore maintain the limiting reference to G06K 1/00.


L/ Regarding
G06K 9/00 (in general)
, SE’s comments regarding ‘computer vision’ and
‘data fusion’ has been addressed by Rapporteur in the above p
aragraph 1, items B and
C.


M/
Regarding
G06K 9/00 (definition statement)
, SE proposes changing paragraphs 3
and 4 as follows:



Image acquisition
specially adapted

for specific patterns, e.g. fingerprint or iris sensors.



Image pre
-
processing
specially
adapted for specific patterns
, e.g. image thinning, noise
filtering.

Rapporteur believes the change
for
paragraph 3 is in compliance with item 64 of the
IPC Guide, and therefore has incorporated the change in the definition statement.

Rapporteur however d
oes not approve the change for paragraph 4 as it narrows the
subject matter. The subject matter of image pre
-
processing in this area does not
necessarily dictate the use for specific patterns exclusively as suggested by SE.

SE questions whether the word ‘
interface’ in the last paragraph under the
definition
statement

section should be ‘interfaces’ instead.
Rapporteur believes the phrase “user
interface” as currently provided in the last paragraph is proper

and therefore has
maintained it
.


N/
Regarding

G06K 9/00 (references relevant to classification in this group), SE makes
comments that the references to G01T 5/02 and G1
0
L 15/00, and
several
references
(
such as A61B 5/117
) as shown in the list ‘examples of places where the subject matter
of this class

is …


are merely informative.
In light of SE’s comments and upon further

8

review,
Rapporteur
believes the references to B25J 9/00; G01M, G01N, G01S, G01T
5/02, G01V, G03F, G06F 17/20, G06F 17/30, G06F 19/00, G06F 21/00,
G06Q,
G08B
13/194, G08G 1/
0
0, and G
10L 15/00 under the
references relevant to classification in
this group

section could be considered informative, and therefore has moved the
respective references to the
informative references

section
.

Rapporteur has updated
the status of informative re
ferences (see Annex 34) to reflect the
status
change

of G01T
5/02 and G10L 15/00
.


Rapporteur however believes
the limiting references to
A61B 5/00, A61B 5/117, B07C

3/10

(changed
from
B07C
for clarity
), G06F 3/01,

G07D 7/20, and G06K 11/00

under
the
refer
ences relevant to classification in this group

section
have overlapping subject
matter with G06K 9/00
, and therefore Rapporteur has maintain
ed

the respective limiting
references.


O/

Regarding
G06K
9/00
(glossary of terms)
, SE

propos
es to modify
the defin
ition of
cluster as “extracting
a common

structure from a set of patterns, i.e. grouping patterns
according to their similarity or closeness”.

Rapporteur however believes it would more
appropriate to define cluster as “extracting
a like

structure from a s
et of patterns, i.e.
grouping patterns according to their similarity or closeness”
, and therefore Rapporteur
has adopted this definition in the
glossary of terms
.



P/ Regarding
G06K 11/00 (definition statement)
,
SE makes remarks that the definition
state
ment is inaccurate since it only covers 11/06. In light of SE’s comments,
Rapporteur has added “
A
utomatic curve followers using an auxiliary scanning pattern”
under
the G06K 11/00 definition statement.

SE makes remarks that the current definition statemen
t for G06K 11/00 is very similar to
G06F 3/033,
with the
particular use of same examples for both areas and thereby
rendering confusion. In view of SE’s comments

and suggestions
, Rapporteur has
deleted
the example “, e.g. hand
-
held sensing devises like li
ght
-
pens, joysticks, mice or
trackballs” under the
definition statement

section
. Rapporteur believes G06K 11/00
covers hand
-
held sensing devices insofar as these devices are not meant as input
devices for digital computers or interaction between user and
computer, instead these
devices are being covered by group G06F 3/033. For clarity, Rapporteur has added a
limiting reference to G06F 3/033 “
Hand
-
held input or output devices for transferring data
to be processed into a form capable of being handled by a
digital computer, e.g. light
-
pens, joysticks, mice, or trackballs”
under the
references relevant to classification in this
group

section of the definition.


Q/ Regarding

G06K 11/00 (relationship between larger subject matter areas)
, SE makes
remarks that

entries in this section should be running text.

SE is correct and
Rapporteur thanks SE for pointing this out. Since there is already a limiting reference to
G06F 3/033 (see above item P), Rapporteur believes it is not necessary to include
a

reference to

G06F 3/00 under the
relationship between large subject matter areas

9

section

of the definition
, and therefore Rapporteur has deleted the reference to G06F
3/00.


R/
Regarding
G06K 13/00 (informative references)
, SE makes comments that the
reference to G11B

15/00 is not consistent with the title of G11B 15/00 in the scheme.
Rapporteur has change the informative reference to G11B 15/00 as “
Driving, starting or
stopping
record carriers

of filamentary or web form
;
Driving both such
record carriers

and
heads
;
Guiding such
record carriers

or containers therefor
;
Control thereof
;
Con
trol
of operating function
”, as suggested by SE.


S/ Regarding
G06K 13/00 (definition statement)
,
SE questions whether the definition
statement is accurate as the statement only covers two
subgroups.
In view of SE’s
comments, Rapporteur has
(1)
added “tra
nsporting of cards between stations”
and
“handling flexible sheets” under “Conveying cards, e.g.”; and (2) added “guiding of
tapes
” and “
recognising end of tape” under “Co
n
veying punched tapes, e.g.”
for the
definition statement
.


T/ Regarding
G06K 15/00
(
d
efinition statement)
,
SE makes remarks that the current
definition statement covers any printing, which is not very helpful. Rapporteur however
believes the definition statement is an accurate reflection of the scope of this group as
the definition state
ment is derived from the subgroup titles (e.g., 15/02 and 15/22).
Rapporteur will therefore maintain the current definition statement.


U/ Regarding
G06K 17/00 (definition statement)
,
SE makes comments that the
statement does not fully cover the subject m
atter of the main group and that the
statement appears broader than the scope of the title which only deals with “effecting
co
-
operative working between equipments”. Rapporteur believes the subject matter
covered by this group in general is broad since
,

a
s the titile indicates, that
it is covered
by two or more of the preceding main groups.
As there are numerous combinations of
the “two or more of the preceding main groups” and w
hile it may be true that the
statement is broader than the scope of the title
,
Rapporteur
believes
the current
definition statement
providing the exemplary coverages as listed
is sufficient
.


V/

Regarding
G06K 19/00 (in general)
,
SE believes that a section for relationship should
be added, in particular that section should contain
the information in the first reference
after the main group title.
In light of SE’s comments, Rapporteur has added
a
relationship
between large subject matter areas

section in the 19/00 definition with
“This
main group does not cover record carriers adapt
ed for controlling specific machines
,

such as in B23Q (details, components, or accessories for machine tools), D03C

(shedding mechanisms; pattern cards or chains; punching of cards; designing patterns)
,
G10F

(automatic musical instruments)
, H04L

(transmiss
ion of digital information);

10

printing, duplicating, marking, or copying processes
B41M,
file cards
B42F,
record
carriers in general
G11B
” provided underneath.




W/ Regarding
G06K 19/00 (definition statement)
,
SE makes remarks that the statement
only c
overs the three two
-
dot groups 19/063, 19/067, and 19/08 and not the entire
scope of the main group. Rapporteur believes the definition statement is sufficient and
therefore will maintain it.


X/ Regarding
G06K 19/00 (references relevant to classification

in this group)
, SE makes
remarks that the reference to B42F 19/00 points the wrong wa
y

as it is a broader place,
and further comments that all other references given in this section should be moved to
the
informative references

section.
Rapporteur has mo
ved the reference to B42F 19/00
to the
informative references

section as suggested by SE
, and the status of informative
references has been updated accordingly (see Annex 34). Regarding the

other
references list in the
references relevant to classificatio
n in this group

section
, however,
Rapporteur believes they

are proper
ly listed in this area

since the respective references
have overlapping subject matter with G06K 19/00. Rapporteur has therefore
maintained these references in the
references relevant to

classification in this group

section.


Y/
Regarding
G06K 21/00 (references relevant to classification in this group)
,
SE makes
comments that this group overlaps with at least 9/00 and 13/00 and clarification is
required between these main groups.

In ligh
t of SE’s comments,
Rapporteur has added
the following limiting references under

the
references relevant to classification in this
group

section:


Detection or correction of errors by rescanning patterns

G06K 9/03

Checking correct operation of card
-
convey
ing mechanisms

G06K 13/06


SE also indicates that the word “handling” in the title is not used in the meaning
standardized in the Guide.
The word “handling” has been defined in the
Guide (page
36) as


dealing with
material

or

objects in any way without intentionally or essentially
altering any property, even temporarily (e.g., without deforming, heating, electrifying),
e.g., transporting, storing, positioning, dispensing, winding, loading”.


Rapporteur
believe
s the use of the

word “handling” in the title of G06K 21/00 is consistent with the
meaning as defined in the Guide
, and
Rapporteur will
therefore maintain it in the title.



Z/
Regarding
“Comment relating to annex 16”
, SE points out that the informative
reference to G01
B is missing in
annex 16,
list

of informative references
.
SE is correct

11

and Rapporteur
thanks SE for point this out. Rapporteur
has updated the status of
informative references (see Annex 34) to include the missing reference to G01B.





[End]