Ex parte Ball

Board of Patent Appeals and Interferences

Mar 30, 2001

08389096 (B.P.A.I. Mar. 30, 2001)

1
The opinion in support of the decision being entered today was
not written for publication and is not binding precedent of
the Board.
Paper No. 11
UNITED STATES PATENT AND TRADEMARK OFFICE
__________
BEFORE THE BOARD OF PATENT APPEALS
AND INTERFERENCES
__________
Ex parte STUART R. BALL
__________
Appeal No. 1998-0752
Application 08/389,096
___________
ON BRIEF
___________
Before KRASS, FLEMING, and GROSS, Administrative Patent
Judges.
FLEMING, Administrative Patent Judge.
DECISION ON APPEAL
This is a decision on appeal from the final rejection of
claims 1-4 and 8-12. Claims 1-4 and 8-17 are pending in the
application; claims 13-17 have been allowed.
The invention relates to methods for recognition of
printed characters, particularly magnetically scanable
characters on bank documents. In one embodiment of the
Appeal No. 1998-0752
Application 08/389,096
2
invention, the method includes scanning a document having such
characters to generate a waveform having a time varying
amplitude (specification, page 15, lines 6-15); sampling the
waveform at a predetermined number of positions (page 15,
lines 21-28); normalizing the samples by dividing each sample
by the computed average magnitude of all samples (page 16,
lines 18-23); determining the difference in amplitude between
each normalized sample and the previous normalized sample to
generate a set of "ratio difference values" (page 17, lines 5-
12); and comparing that set of ratio difference values to
stored sets of ratio difference values, the closest match
being selected as the recognized character (page 17, lines 12-
16). In another embodiment of the invention, the method uses
the numerical difference between normalized samples to
generate a set of identifiers that determine whether the
waveform has a rising slope, falling slope, or no change in
slope from one sample to the next, the set of identifiers
being compared with stored sets of identifiers in order to
make a character recognition decision (page 18, lines 11-28).
Independent claims 1 and 9 are reproduced as follows:
Appeal No. 1998-0752
Application 08/389,096
3
1. A method for recognizing one or more characters on a
document and the like comprising the steps of:
scanning said document with a device which is operable to
generate a signal having a time varying amplitude represented
by a waveform which is peculiar to the shape of a character
being scanned;
determining the amplitude of said waveform at a
predetermined number of positions on a time scale of said
waveform, representing one dimension of said character, to
provide a set of sample values of the amplitude of said
waveform;
calculating the amplitude ratio value of a predetermined
number of samples of a set by dividing each sample magnitude
by the average magnitude of the predetermined number of
samples of the set;
determining the difference between selected ones of said
sample amplitude ratio values with respect to an adjacent
sample amplitude ratio value to generate a set of ratio
difference values; and
comparing said set of ratio difference values to
predetermined sets of ratio difference values, each of said
predetermined sets of ratio difference values representing the
identity of a character, to recognize the character being
scanned.
9. A method for recognizing one or more characters on a
document and the like comprising the steps of:
scanning said document with a device which is operable to
generate a signal having a time varying amplitude waveform
which is peculiar to the shape of a character being scanned;
determining the amplitude of said waveform at a
predetermined number of positions on a time scale of said
Appeal No. 1998-0752
Application 08/389,096
4
waveform, representing one dimension of said character, to
provide a set of sample values of the amplitude of said
waveform;
calculating the amplitude ratio value of each sample of a
set by dividing each sample magnitude by the average sample
magnitude of the set;
determining the numerical difference between selected
ones of said sample amplitude ratio values with respect to an
adjacent sample amplitude ratio value to generate a set of
identifiers which determine whether said waveform has a rising
slope, a falling slope, or no change in slope between
respective ones of said selected ones of said sample amplitude
ratio values; and
comparing said set of identifiers with plural
predetermined sets of identifiers, each representing the
identity of a character, to recognize the character being
scanned.
The Examiner relies on the following references:
Toyama 4,399,553 Aug. 16,
1983
Kao 4,245,211 Jan. 13,
1981
Roberts 4,032,887 Jun. 28,
1977
Appellant's admitted Prior Art ("AAPA")
Claims 1, 2, and 8-10 stand rejected under 35 U.S.C. §
103 as being unpatentable over Toyama in view of Kao and AAPA.
Claims 3, 4, 11, and 12 stand rejected under 35 U.S.C. § 103
as being unpatentable over Toyama in view of Kao, AAPA, and
Roberts.
Appeal No. 1998-0752
Application 08/389,096
5
Rather than repeat the arguments of Appellant or the
Examiner, we make reference to the brief and the answer for
the details thereof.
OPINION
After a careful review of the evidence before us, we
agree with the Examiner that claims 9, 11, and 12 are properly
rejected under 35 U.S.C. § 103. Thus, we will sustain the
rejection of these claims; but we will reverse the rejection
of the remaining claims on appeal for the reasons set forth
infra.
It is the burden of the Examiner to establish why one
having ordinary skill in the art would have been led to the
claimed invention by the reasonable teachings or suggestions
found in the prior art, or by a reasonable inference to the
artisan contained in such teachings or suggestions. In re
Sernaker, 702 F.2d 989, 995, 217 USPQ 1, 6 (Fed. Cir. 1983).
In addition, the Federal Circuit states that "[t]he mere fact
that the prior art may be modified in the manner suggested by
the Examiner does not make the modification obvious unless the
prior art suggested the desirability of the modification." In
Appeal No. 1998-0752
Application 08/389,096
6
re Fritch, 972 F.2d 1260, 1266 n.14, 23 USPQ2d 1780, 1783-84
n.14 (Fed. Cir. 1992), citing In re Gordon, 733 F.2d 900, 902,
221 USPQ 1125, 1127 (Fed. Cir. 1984).
"Additionally, when determining obviousness, the claimed
invention should be considered as a whole; there is no legally
recognizable 'heart' of the invention." Para-Ordnance Mfg. v.
SGS Importers Int’l, Inc., 73 F.3d 1085, 1087, 37 USPQ2d 1237,
1239 (Fed. Cir. 1995), cert. denied, 117 S. Ct. 80 (1996),
citing W. L. Gore & Assocs., Inc. v. Garlock, Inc., 721 F.2d
1540, 1548, 220 USPQ 303, 309 (Fed. Cir. 1983), cert. denied,
469 U.S. 851 (1984). In addition, the Federal Circuit reasons
in Para-Ordnance Mfg., 73 F.3d at 1087-88, 37 USPQ2d at 1239-
40, that for the determination of obviousness, the court must
answer whether one of ordinary skill in the art who sets out
to solve the problem, and who had before him in his workshop
the prior art, would have been reasonably expected to use the
solution that is claimed by the Appellant.
On page 12 of the Brief, Appellant argues that the
combination of Toyama, Kao, and AAPA fails to teach the
invention of claim 9 because the combination allegedly lacks a
Appeal No. 1998-0752
Application 08/389,096
7
teaching of determining the numerical difference between
selected ones of the sample amplitude ratio values with
respect to an adjacent sample amplitude ratio value to
generate a set of identifiers which determine whether the
waveform has a rising slope, a falling slope or no change in
slope and then comparing the set of identifiers with plural
predetermined sets of identifiers, each representing the
identity of a character.
With respect to claim 9, we find that Toyama teaches a
method for recognizing (magnetic ink) characters on a
document, comprising scanning the document with a device
operable to generate a signal with an amplitude waveform
peculiar to the shape of the character being scanned (column
3, lines 3-29); determining the amplitude of the waveform at a
predetermined number of positions (column 3, lines 50-60;
sample and hold circuit 17); determining the numerical
difference between adjacent samples to generate a set of
identifiers that indicate rising slope, falling slope, or no
slope change (column 4, lines 25-56; comparators 20, 21, flip-
flops 22, 23); and comparing the set of identifiers with
Appeal No. 1998-0752
Application 08/389,096
8
plural predetermined sets of identifiers in order to recognize
the character being scanned (column 4, line 57 to column 5,
line 19). Toyama lacks a teaching of normalizing the sampled
amplitudes, i.e., "calculating the amplitude ratio value of
each sample of a set by dividing each sample magnitude by the
average sample magnitude of the set."
Kao suggests, in the context of reading magnetic ink
characters, normalizing sampled amplitudes by dividing each
sample value by the largest sample found (column 5, line 55 to
column 6, line 15). Appellant admits at page 3, lines 9-22 of
the specification that "recently, methods have been developed"
whereby "[t]he series of digital words, representing the full
character width, are summed to create an average and each
individual word is divided by the average to generate a new
digital word that represents the ratio of the individual
digital words to the average."
We find that those skilled in the art having the
teachings of Toyama, Kao, and AAPA before them would have
normalized the sample amplitudes by dividing each amplitude by
the average of all samples, because Kao suggests that
Appeal No. 1998-0752
Application 08/389,096
9
normalization "makes all characters the same size" (column 5,
lines 58-59). Kao recognizes that "waveform amplitude is a
function of the ink signal strength" (column 3, lines 56-57).
Appellant admits at page 2 of the specification that many
problems may cause character reading devices to produce signal
defects. The person having ordinary skill in the art would
have recognized that Kao's suggestion to make all characters
the same "size" (i.e., amplitude) would solve the problem of
different documents producing signals of different amplitudes.
Therefore, we find that it would have been obvious to one
skilled in the art in view of the teachings of Toyama, Kao,
and AAPA to calculate the amplitude ratio value of each sample
of a set by dividing each sample magnitude by the average
magnitude, then generate a set of identifiers, then compare
each set of identifiers with predetermined sets of identifiers
to recognize a character being scanned, as recited in claim 9.
On pages 12 and 13 of the Brief, Appellant urges that
claims 11 and 12 are patentable for the reasons set forth in
support of the patentability of claims 3 and 4, respectively.
Appellant does not discuss the limitations of claim 9, from
Appeal No. 1998-0752
Application 08/389,096
10
which claims 11 and 12 depend, that are not present in claim
1, from which claims 3 and 4 depend. Specifically, as noted
supra, we find that claim 9's limitation of "generating a set
of identifiers" which signify rising slope, falling slope, or
no slope is taught by Toyama. Because that limitation is not
present in claim 1, we are not persuaded that claims 11 and 12
should be patentable for the same reasons that claims 3 and 4,
respectively, are patentable.
With respect to claims 11 and 12, we find that Kao
teaches character recognition "by finding a correlation
between the normalized waveshape and one of the defined
character waveform patterns stored in ROM" (column 6, lines
16-19). Kao thus teaches comparing amplitude ratios (i.e.,
normalized amplitudes) of the samples of each set with
predetermined sets of amplitude ratios, in order to identify
which character has been read. We further find that Roberts
teaches applying plural pattern recognition techniques to a
set of input data, and selecting an output from among the
recognition methods (see column 2, lines 1-12, and Figure 3).
We find that those skilled in the art, having the
Appeal No. 1998-0752
Application 08/389,096
11
teachings of Toyama, Kao, AAPA and Roberts before them, would
have included the ability to recognize a character by
comparing a set of normalized amplitudes with predetermined
sets of normalized amplitudes, and selecting the set with the
highest correlation as the identified character, as taught by
Kao, because Kao suggests that normalized MICR waveform
amplitude samples may be effectively compared with stored
amplitude samples to recognize printed characters. We further
find that those skilled in the art would have modified the
combination of claim 9 supra to perform simultaneously (at
least) two methods of character recognition, and arbitrate
between the results, because Roberts suggests that the use of
plural recognition systems reduces the number of recognition
errors produced, and makes the system more robust and able to
respond to a variety of inputs (see column 1, lines 65-68).
Therefore, we find that it would have been obvious to modify
the combination of Toyama, Kao, and AAPA, expressed supra with
respect to claim 9, to provide plural recognition methods,
including comparing a set of normalized amplitudes with
predetermined sets, as recited in claim 11, and choosing
Appeal No. 1998-0752
Application 08/389,096
12
between the results of those recognition methods, as recited
in claim 12.
On pages 8 and 9 of the Brief, Appellant argues that the
combination of Toyama, Kao, and AAPA fails to suggest the
invention recited in claim 1, because none of the references
disclose or suggest the step of determining the differences
between adjacent (normalized) samples to generate a set of
ratio
difference values, and them comparing such a set with
predetermined sets of ratio difference values to recognize a
character being scanned.
Upon a careful review of Toyama, Kao, and AAPA, we fail
to find that any reference or combination of references
teaches determining the difference between selected sample
amplitude ratio values and adjacent sample amplitude ratio
values to generate a set of ratio difference values, followed
by comparing that set of ratio difference values with stored
ratio difference value sets to determine a recognized
Appeal No. 1998-0752
Application 08/389,096
13
character, as is recited in claim 1. The system of
comparators and flip-flops used in Toyama serves to produce a
pair of (binary) identifiers that signal whether the slope of
the amplitude curve is rising, falling, or steady. Toyama,
however, does not contemplate generating a set of (numerical)
ratio difference values, as illustrated in Appellant's Figure
3C, to be compared with predetermined sets of ratio difference
values. Kao teaches a MICR waveform analyzer. As noted
supra, however, Kao teaches comparison of
normalized amplitudes to store sets of normalized amplitudes.
Kao does not compute numerical differences between adjacent
samples.
Thus, we fail to find that the combination proposed by
the Examiner would have resulted in the claimed invention.
Claims 2-4 depend from claim 1; therefore, the rejection
of claims 2-4 is reversed for the same reasons expressed supra
with regard to claim 1. Independent claim 8 contains
limitations parallel to those contained in claim 1, i.e.,
Appeal No. 1998-0752
Application 08/389,096
14
determining the numerical differences between adjacent sample
amplitude ratio values, and comparing the set of ratio
difference values to previously stored sets to determine the
"event" recognized. Therefore, we find that the prior art
relied upon by the Examiner fails to teach the limitations of
claim 8, for the same reasons specified with respect to claim
1. Thus, we will not sustain the rejection of claims 1-4 and
8 under 35 U.S.C. § 103.
On page 12 of the Brief, Appellant argues that claim 10
is patentable because the step of setting threshold values of
differences between adjacent sample amplitude ratio values, in
order to identify whether the slope is rising, falling, or
unchanged, is not taught by the combination of Toyama, Kao,
and AAPA advanced by the Examiner.
Upon a careful review of the references applied by the
Examiner, we fail to find that any one of Toyama, Kao, or AAPA
teach the setting of threshold values of the difference
between normalized sample amplitudes which define whether
their slope is rising, falling, or unchanging. Toyama merely
teaches a two comparator system, in which one comparator
Appeal No. 1998-0752
Application 08/389,096
15
output goes high to indicate a rising slope, the other
comparator output goes high to indicate a falling slope;
should both outputs be low, the result is interpreted as no
change (Figures 4 and 6). Toyama does not teach the ability
to set a numerical difference between adjacent normalized
samples that will be defined as the threshold between rising
slope and no slope, or between no slope and falling slope.
Thus, we will not sustain the rejection of claim 10 under 35
U.S.C. § 103.
In view of the foregoing, the decision of the Examiner
rejecting claims 9, 11, and 12 under 35 U.S.C. § 103 is
affirmed; the decision of the Examiner rejecting claims 1-4,
8, and 10 under U.S.C. § 103, however, is reversed.
Appeal No. 1998-0752
Application 08/389,096
16
No time period for taking any subsequent action in
connection with this appeal may be extended under 37 CFR
§ 1.136(a).
AFFIRMED-IN-PART
ERROL A. KRASS )
Administrative Patent Judge )
)
)
) BOARD OF PATENT
MICHAEL R. FLEMING )
Administrative Patent Judge ) APPEALS AND
)
) INTERFERENCES
)
ANITA PELLMAN GROSS )
Administrative Patent Judge )
Appeal No. 1998-0752
Application 08/389,096
17
MRF:pgg
Kenneth R. Glaser
Akin Gump Strauss Hauer & Feld
1700 Pacific Avenue
Suite 4100
Dallas, TX 75201