Ex Parte Aldrich et al

Patent Trial and Appeal Board

Sep 10, 2013

10812626 (P.T.A.B. Sep. 10, 2013)

UNITED STATES PATENT AND TRADEMARK OFFICE
__________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
__________

Ex parte BRADLEY C. ALDRICH, MOINUL H. KHAN,
and KAYLA L. CHALMERS1
__________

Appeal 2011-005464
Application 10/812,626
Technology Center 2600
__________

Before ERIC GRIMES, JEFFREY N. FREDMAN, and
ERICA A. FRANKLIN, Administrative Patent Judges.

GRIMES, Administrative Patent Judge.

DECISION ON APPEAL
This is an appeal under 35 U.S.C. § 134 involving claims related to
image processing, which have been rejected for obviousness. We have
jurisdiction under 35 U.S.C. § 6(b).
We reverse.

1 Appellants identify the Real Party in Interest as Marvell International Ltd.
(Appeal Br. 3).
Appeal 2011-005464
Application 10/812,626

2
STATEMENT OF THE CASE
The Specification states that image data processing operations, such as
gamma correction or illumination correction, “may be implemented using a
transfer function to describe the appropriate adjustment. A transfer function
is merely a function that maps input values to output values according to a
formula or curve.” (Spec. 2, ¶ 3.) The Specification states that
implementing transfer functions as look-up tables (LUTs) increases
processing speed but takes up a lot of memory (id. at 2-3, ¶ 4).
Storing only sample outputs of the transfer function, on the other
hand, reduces the memory required but compromises accuracy and
flexibility (id. at 3, ¶ 4). Some of the accuracy lost by sampling can be
regained by performing interpolation (id. at 8, ¶ 22) but when the curve
representing the transfer function “is not straight or close to straight, i.e., has
high curvature, interpolation cannot recapture some, or much, of the
accuracy lost to sampling” (id. at 9, ¶ 23). The Specification discloses that
[m]any real life transfer functions have both high-curvature and
low-curvature portions, as shown in Figure 3. Since LUTs 124
are fixed in size, they can only store a certain number of
samples. Since even a relatively few samples can accurately
reflect the low-curvature regions of the transfer function, when
used with interpolation, it can help accuracy to reallocate more
of the fixed number of samples to the high-curvature regions of
the transfer function.
(Id. at 9, ¶ 24.)
Claims 1-18, 24-27, and 32-34 are on appeal. Claims 1 and 24 are
illustrative and read as follows (emphasis added):
1. An image processing device comprising:
a look-up table (LUT) storing sample outputs from an output
range of a transfer function, wherein the transfer function maps sample
Appeal 2011-005464
Application 10/812,626

3
inputs from an input range of the transfer function to the sample outputs, and
wherein, based on a curvature of the transfer function, the sample inputs are
distributed so that more sample inputs are associated with a first region of
the transfer function than a second region of the transfer function; and
an address module to calculate an index into the LUT based on
image data.

24. A method comprising:
receiving image data, the image data being input for a transfer
function, the transfer function mapping an input range to an output range;
using a first section of the received image data to identify a
region of the input range of the transfer function to which the received
image data belongs;
selecting a second section of the received image data based on
the identified region;
addressing an entry of a look-up table (LUT) using the first and
second sections of the image data; and
calculating a transferred image data by using the addressed
entry and a residual section of the image data.

Claims 11 and 32, the only other independent claims, include
limitations similar to those of claims 1 and 24, respectively.
The claims stand rejected under 35 U.S.C. § 103(a) as follows:
• Claims 1-11, 24-27, and 32-34 based on Reitan2 and Leffel3
(Answer 3);
• Claim 11 based on Reitan and Takane4 (Answer 9); and
• Claims 12-18 based on Reitan, Takane, and Leffel (Answer 11).
In each of the rejections on appeal, the Examiner relies on Reitan as
disclosing the claim limitations related to a transfer function (Answer 3-11).

2 Reitan, 5,600,574, issued Feb. 4, 1997.
3 Leffel, 2005/0057303 A1, published Mar. 17, 2005.
4 Takane, 2002/0030751 A1, published Mar. 14, 2002.
Appeal 2011-005464
Application 10/812,626

4
The Examiner points to specific passages in Reitan’s columns 16, 17, 18, 21,
and 22, as well as Figures 9-11, as support for this finding (Answer 3-4).
The Examiner relies on Leffel and Takane for disclosure of an address
module (claim 1) and a digital camera (claim 11), respectively (id. at 4, 10).
Appellants argue that “each of the[ ] cited portions is absent of any
teaching or suggestion whatsoever that sample inputs are distributed so that
more sample inputs are associated with a first region of the transfer function
than a second region of the transfer function specifically based on the
curvature of the transfer function” (Appeal Br. 10-11).
We agree with Appellants that the Examiner has not provided an
adequate basis to support the finding that Reitan discloses a look-up table
storing sample outputs “wherein, based on a curvature of the transfer
function, the sample inputs are distributed so that more sample inputs are
associated with a first region of the transfer function than a second region of
the transfer function,” as required by claim 1.
Reitan discloses an “apparatus and method for automatically testing
the functional components of an electronic digital imaging system” (Reitan,
col. 2, ll. 49-51). Reitan describes, among other things, testing for pixel
value integrity using a “sufficient number of sample points on reference film
405 . . . to reveal any nonlinearities over small ranges of input values” (id. at
col. 16, ll. 12-14); a method of “measuring the ‘laser beam wobble’ in a
laser-based film digitizer” by measuring “the total deviation from
straightness in the sample image due to this effect” (id. at col. 16, ll. 36-53);
and analyzing digital sample images “by the image quality metrics software
to locate specific regions of interest and to compute measurements for each
Appeal 2011-005464
Application 10/812,626

5
region of interest” (id. at col. 17, ll. 31-34). The Examiner has not explained
why any of these disclosures, which are cited in the Answer, would have
been understood by a skilled artisan to describe distributing sample inputs
associated with a transfer function based on the curvature of the transfer
function.
Some of the other passages cited by the Examiner do expressly
mention transfer functions. For example, Reitan describes a rotation-
independent modulation transfer function (id. at col. 18, ll. 30-31). Reitan
also states that “Look Up Tables are utilized within the electronic imaging
system of the present invention to perform conversion of an incoming pixel
data set to an outgoing pixel data set so as to achieve a desired transfer
function for a particular system component” (id. at col. 21, ll. 57-60). Reitan
states that in a preferred embodiment, “LUT’s are used to transform pixel
quantities. There are three LUT’s in the basic system.” (Id. at col. 22,
ll. 6-8.)
These passages, while referring generally to transfer functions and
look-up tables, do not expressly describe distribution of the sample inputs
based on any particular characteristic of the transfer function. Specifically,
they do not state that “based on a curvature of the transfer function, the
sample inputs are distributed so that more sample inputs are associated with
a first region of the transfer function than a second region of the transfer
function,” as required by claim 1. In addition, Examiner has not explained
why the cited passages would have been understood by those skilled in the
art to describe the disputed limitation. We conclude that the Examiner has
Appeal 2011-005464
Application 10/812,626

6
not established that this limitation of the claimed invention would have been
obvious to a person of ordinary skill in the art based on Reitan.
With regard to claim 24, Appellants argue that it includes limitations
similar to those of claim 1 and is therefore allowable for the same reasons
(Appeal Br. 13).
We agree with Appellants that the Examiner has not persuasively
shown that the method of claim 24 would have been obvious based on the
cited references. Claim 24 requires “using a first section of the received
image data to identify a region of the input range of the transfer function to
which the received image data belongs [and] selecting a second section of
the received image data based on the identified region” (claim 24).
The Examiner equates the first and second sections of the received
image data that are recited in the claim to “regions [that] can be identified
based on the interest of the process. See Column 17, Lines 28-45.” (Answer
7.) The cited portion of Reitan states that “digital sample images . . . are
analyzed by the image quality metrics software to locate specific regions of
interest and to compute measurements for each region of interest” (Reitan,
col. 17, ll. 31-34). That is, the cited passage refers selecting regions of an
image for analysis.
Claim 24, however, refers to first and second sections “of the received
image data.” The Specification makes clear that the claim language refers to
sections of the data itself, not data associated with regions of an image. The
Specification describes dividing the curve of the transfer function into
different regions, such as quartiles, which can be low curvature regions or
higher-curvature regions (Spec. 9, ¶ 25); if there are regions identified as
Appeal 2011-005464
Application 10/812,626

7
high curvature, more sample inputs are allocated to those regions (id. at 10,
¶ 27).
The Specification states that the image data are divided up into three
fields, referred to as the Q (quadrant5), C (coarseness), and R (residue)
fields; “these fields can be referred to as first, second, and third parts or
sections of the image data, respectively” (id. at 11, ¶ 30, emphasis added).
The Specification states that the Q field is made up of two bits of the image
data, while the C field can have either two or three bits, depending on how
many samples were allocated to a particular quadrant/quartile (id. at 11,
¶ 31). The Q field and the C field are then used to identify an entry in the
LUT (id. at 12, ¶ 34).
Thus, when the claim language is read in light of the Specification, the
recited first and second sections of the received image data refer to different
parts of the image data itself, not image data relating to different sections of
an image. The Examiner has not shown that claim 24, when interpreted in a
manner consistent with the Specification, would have been obvious based on
the cited references.
As noted above, claims 11 and 32 include limitations similar to claims
1 and 24, respectively. We therefore reverse the rejection of all the claims
on appeal.

5 The cited passage describes the embodiment shown in Figure 6, which
refers to “quartiles” rather than “quadrants” (Spec. Fig. 6: “quartile pointers
130”).
Appeal 2011-005464
Application 10/812,626

8
SUMMARY
We reverse the rejection all of the rejections on appeal.

REVERSED

cdc