DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Notice to Applicants
This action is in response to the amendments and remarks filed on 08/25/2026.
Claims 1-2, 4-5, 7-9, and 11 are pending.
Corrective Actions by Applicant
Claims 1-2, 4-5, 7, 9, and 11 have been amended.
Claims 3, 6, 10, and 12 have been canceled.
Response to Arguments
The examiner has fully considered Applicant’s presented arguments.
On page 6 of the remarks, Applicant argues that the claimed search and matching windows would be understood by those of ordinary skill in the art and their depiction in the drawings is not needed. This is persuasive, based on the discussion in the interview of 08/20/2026. More specifically, the removal of the special case of the 81 matching blocks from claims 2, 5, 9, and 11 overcomes the objection to the drawings, as otherwise the claimed search and matching windows are now understandable to one of ordinary skill in the art as part of the NLMeans algorithm. The drawing objections have thus been withdrawn.
On page 6 of the remarks, Applicant argues that the amendments to claims 2, 5, 9, and 11 overcome the 35 U.S.C. 112(a) and 112(b) rejections of the same. This is persuasive, based on the removal of the special case of the 81 matching blocks. All 112(a) and 112(b) rejections have thus been withdrawn.
On page 7 of the remarks, Applicant argues that all claims as amended are eligible under 35 U.S.C. 101. This is persuasive, based on Applicant’s implementation of the examiner’s suggested amendments to overcome the 101 rejections. All 101 rejections are thus withdrawn.
On pages 8-10 of the remarks, Applicant argues that Rodrigues is not in the field of NLMeans-based denoising, and Rodrigues’s equation (9) is not the same as the claimed formula.
While the examiner agrees that Rodrigues does not use the claimed equation for a denoising application, Rodrigues still specifies that their equation 9 is usable as a replacement for the standard Euclidean distance for image processing applications (see the previously-cited table 1 and section 3 of Rodrigues). The Euclidean distance is what Buades uses to determine the similarities between the matching and central blocks for NLMeans denoising, hence the support for the combination by replacing Buades’s Euclidean distance with Rodrigues’s equation 9.
However, the examiner agrees with Applicant’s argument that Rodrigues’s equation 9 does not fully disclose the claimed equation. Specifically, the claimed equation uses two weights “(1-a)” and “a” that multiply the Manhattan and Chebyshev distances, respectively. Rodrigues, in comparison, uses two general weights “w1” and “w2”; although the first paragraph of Rodrigues section 3.1 describes the inverse relationship between these weights (“As w1 increases in regard to w2, the distance becomes more like Minkowski. Contrariwise, it converges towards Chebyshev”), and the following paragraph specifies that w1, w2 > 0, the weights are still not disclosed to have the exact relationship of “(1-a)” and “a” as claimed.
Thus, all previous 103 rejections have been withdrawn. However, the claim rejections necessitate new 103 rejections in further view of Wong below.
PNG
media_image1.png
301
512
media_image1.png
Greyscale
Rodrigues Table 1 (see below argument)
On pages 10-11 of the remarks, Applicant argues that modifying Buades with Rodrigues requires impermissible hindsight. The examiner respectfully disagrees. As also discussed in the above argument response, Rodrigues provides clear and compelling motivation to one of ordinary skill in the art to replace the Euclidean distance, such as the distance used by Buades, with Rodrigues’s equation 9, as doing so is much more computationally efficient.
See Rodrigues Table 1, which compares the processing times for each individual distance formula to Rodrigues’s; also see section 3, “Methodology”, which discusses the performance improvement at various points (on page 68, left column, “Any distance other than Chebyshev and Manhattan requires much more effort. In the Euclidean distance, a quantization/approximation is required. The distances can be pre-computed and stored in a kernel or processed in real time. Storing large amounts of data in kernels consume a significant amount of memory when computing large distances. Accessing this data also impacts negatively on the overall processing performance”; on page 68, right column, “On the other hand, using the Euclidean distance implies a huge performance burden, where its implementation is approximately 368 times slower than Chebyshev for D = 2500 , and it gets even worse as D increases. This work proposes an intermediate distance between Chebyshev and Euclidean that adds information, as opposed to Manhattan, and is far more efficient when it comes to neighbourhood iterations.”).
On pages 12-13 of the remarks, Applicant argues that a person of ordinary skill in the art would not have increased the cost of implementing the features of the claimed invention without recognizing the problem identified by the present inventors or recognizing any other reason for employing the Si formula in the manner claimed.
The examiner respectfully disagrees. The present case differs from the case law cited by Applicant, In re Omeprazole Patent Litigation, 536 F.3d 1361, 87 USPQ2d 1865 (Fed. Cir. 2008). In that case, the claimed invention had been rendered not obvious because, although the prior art did disclose coated omeprazole tablets and general secondary subcoatings, the problem solved by combining these prior art references, i.e. the degradation of the active drug ingredient by an enteric coating, was not disclosed or recognized by the prior art. Thus, the district court concluded that one of ordinary skill in the art would have had no reason to perform the work to combine the prior art to solve a problem that was unrecognized by the prior art. The MPEP expands on this by describing how, if this adverse interaction between the active agent and coating had been known, it might well have been obvious to use a subcoating by combining the prior art. See MPEP 2143.I.A, Example 3, for both the description of this case law and this expansion.
In the present case, Rodrigues actively recognizes the problem with the prior art, i.e. the poor performance of the Euclidean distance as disclosed by Buades, and offers a solution by replacing Euclidean distance with their equation 9.
The examiner further argues that one of ordinary skill in the art would observe Rodrigues’s description of the poor computational performance of the Euclidean distance, and recognize that that low performance would be a problem in any computational system, especially integrated circuits, of which FPGAs are a species, or more broadly embedded systems as a whole, which are well-known in the art to lack high processing power.
Claim Rejections – 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-2, 4-5, 7-9, and 11 are rejected under 35 U.S.C. 103 as being unpatentable over Buades et al. (“A non-local algorithm for image denoising”, IEEE CPVR paper, 25 July 2005) in view of Rodrigues (“Combining Minkowski and Chebyshev: New distance proposal and survey of distance metrics using k-nearest neighbors classifier”, Pattern Recognition Letters paper, 15 July 2018) and Wong et al. (U.S. Publ. US-2016/0314379-A1).
Regarding claim 1, Buades discloses a simplified method for denoising (see section 3, “NL-means algorithm”) an infrared image (see section 1, “Introduction”, where the method is applicable to any digital image, including infrared) by using non-local means (NLMeans) executed on a field programmable gate array (FPGA) (the method is performable on a computerized system such as a FPGA; paragraphs 0003-0005 of Applicant's originally-filed specification indicate that the standard NLMeans method is performable on FPGAs, albeit with reduced performance), comprising:
step S1: determining a size of a search window and a size of a matching window (see section 3, where search windows Nk denote square neighborhoods/matching windows of a fixed size and centered at a pixel k; see section 5, "Discussion and experimentation", where the search window containing all matching windows can be set to 21-by-21 pixels, and the similarity square neighborhoods/matching windows can be set to 7-by-7 pixels);
step S2: (see section 3, where the similarity between each central block Ni and matching block Nj is measured as a decreasing function of the weighted Euclidean distance);
and step S3: calculating a weight of each matching block based on the similarity (see figures 1-2 and section 3, where the weighting function w(i, j) defines the weights between windows i and j, where more similar gray levels have higher weights),
and calculating a current denoised image based on each matching block and the weight (Section 3 specifies that the denoised/estimated pixel value “NL[v][i]” is computed as the weighted average of all pixels in the image; figures 3 and 5 show example denoised images),
S
i
=
1
-
a
∑
a
i
-
b
i
+
a
max
a
i
-
b
i
Buades fails to disclose the limitations indicated via strikethrough above. More specifically, Buades does calculate a similarity between each matching and central block using the Euclidean distance, just not via a linear superposition on a Manhattan distance and Chebyshev distance as also represented by the claimed equation.
PNG
media_image2.png
409
635
media_image2.png
Greyscale
Rodrigues Equation 9
Pertaining to the same field of endeavor, Rodrigues discloses performing linear superposition on a Manhattan distance and a Chebyshev distance (first see section 2, "Literature review", where the Minkowski distance formula, dp (x, y), imitates the Euclidean distance when p=2, Manhattan distance when p=1, and Chebyshev distance when p approaches positive infinity; then see section 3.1, "Distance Definition", where the proposed distance d w1, w2, p (x,y) is a weighted Minkowski distance plus a weighted Chebyshev distance; thus, when p=1, the formula is a linear superposition / combination of the Manhattan and Chebyshev distances; also see figure 1(l), where the illustrated p=1 case mirrors the octagonal shape of figure 1 of the present application);
wherein in the step S2, a following formula is adopted:
S
i
=
1
-
a
∑
a
i
-
b
i
+
a
max
a
i
-
b
i
(see section 3.1, "Distance Definition", equation 9, d w1, w2, p (x, y); note again that the first term imitates the Manhattan distance when p=1),
wherein Si represents the similarity (the left-hand side of equation 9 is regarded as the distance/similarity metric); ai and bi represents a pixel value of an image block; and a subscript i represents a same position in two image blocks (analogous to xi and yi in equation 9, respectively).
Buades and Rodrigues are considered analogous art, as they are both directed to computing distances for image processing applications. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have integrated the teachings of Rodrigues into Buades by replacing Buade's Euclidean distance with equation 9 because the linear superposition of Manhattan and Chebyshev is much more computationally efficient than the standard Euclidean distance (see Rodrigues table 1 and section 3, “Methodology).
Buades in view of Rodrigues fails to further disclose a represents a parameter. More specifically, the weights w1 and w2 of Rodrigues’s equation 9 fill the same role as in the present equation's (1-a) and a, as both sets of weights multiply the Manhattan and Chebyshev distances by a weighting factor to create a weighted sum; Section 3.1 specifies that these weights merely need to satisfy w1, w2 > 0; however, they are not disclosed to have the specific structure of (1-a) and a as claimed, which functionally represents the weights having a sum of 1.
Pertaining to the same field of endeavor, Wong discloses weights of (1-a) and a, where a represents a parameter (the broadest reasonable interpretation of this weighting structure includes a convex combination, which is a linear combination / superposition where all weights are normalized to sum up to 1; see for clarity figure 1B and paragraphs 0032-0033, where similarity scores can be determined between various image patches; paragraphs 0070-0071 specify that this can be applied to image denoising, effectively disclosing an NLMeans algorithm similar to Buades; then see figure 2A and paragraphs 0038, 0051, where the patch similarity scores of each image channel are summed as a linear combination / superposition, with each weight w1 multiplying each distance d1; finally see paragraphs 0022-0023, 0046, 0051, which specify that the weights can be normalized to sum up to 1 to compute a convex combination of the distance values; in combination with Rodrigues, two weights that sum to 1 can be mathematically represented by the claimed structure of (1-a) and a).
Buades and Wong are considered analogous art, as they are both directed to patch distance calculation for denoising. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have integrated the teachings of Wong into Buades and Rodrigues by using the convex combination structure for Rodrigues's weights, which would predictably succeed as Rodrigues's weights merely need to satisfy w1, w2 > 0 (see Rodrigues Section 3.1), by normalizing the weights because said weights represent the relative importance of signal information from each source and thus need to be weighted accordingly (see Wong paragraph 0022).
Regarding claim 2, Buades in view of Rodrigues and Wong discloses wherein the step S1 comprises: determining the size of the search window as N*N and the size of the matching window as d*d (See Buades citations to step S1 of claim 1 above, where the search window containing all matching windows can be set to 21-by-21 pixels, and the similarity square neighborhoods/matching windows can be set to 7-by-7 pixels),
wherein N=11 and d=3 (Buades Section 5 specifies an N value of 21 and a d value of 7; one could easily lower the values of these result-effective variables to 11 and 3, respectively, to determine a balance between accuracy and processing complexity).
Regarding claim 4, Buades discloses a simplified field programmable gate array (FPGA) (the method is performable on a computerized system such as a FPGA; paragraphs 0003-0005 of Applicant's originally-filed specification indicate that the standard NLMeans method is performable on FPGAs, albeit with reduced performance) for denoising an infrared image by using NLMeans (see section 3), comprising:
a first module configured to determine a size of a search window and a size of a matching window (same citations apply from step S1 of claim 1 above);
a second module configured (same citations apply from step S2 of claim 1 above);
and a third module configured to calculate a weight of each matching block based on the similarity, and calculate a current denoised image based on each matching block and the weight (same citations apply from step S3 of claim 1 above),
S
i
=
1
-
a
∑
a
i
-
b
i
+
a
max
a
i
-
b
i
Buades fails to disclose the limitations indicated via strikethrough above.
Rodrigues and Wong are relied upon to teach these same missing limitations as in claim 1 above, with the same motivations to combine as in claim 1 above. Buades in view of Rodrigues and Wong therefore discloses claim 4 as applied to claim 1 above.
Regarding claim 5, Buades in view of Rodrigues and Wong discloses claim 5 as applied to claim 2 above.
Regarding claim 7, Buades in view of Rodrigues and Wong discloses a non-transitory computer-readable storage medium storing a computer program, wherein the computer program is executed by a processor to perform steps of the simplified method for denoising the infrared image by using NLMeans based on the FPGA according to claim 1 (Buades's method is performable on a computerized system such as a FPGA; paragraphs 0003-0005 of Applicant's originally-filed specification indicate that the standard NLMeans method is performable on FPGAs, albeit with reduced performance, which have computer-readable storage mediums).
Regarding claim 8, Buades in view of Rodrigues and Wong discloses an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable in the processor, wherein the computer program is executed by the processor to perform steps of the simplified method for denoising the infrared image by using NLMeans based on the FPGA according to claim 1 (Buades's method is performable on a computerized system such as a FPGA; paragraphs 0003-0005 of Applicant's originally-filed specification indicate that the standard NLMeans method is performable on FPGAs, albeit with reduced performance, which have memories, processors, and computer programs stored in said memories).
Regarding claim 9, Buades in view of Rodrigues and Wong discloses claim 9 as applied to claim 2 above.
Regarding claim 11, Buades in view of Rodrigues and Wong discloses claim 11 as applied to claim 2 above.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Contact Information
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NICHOLAS JOHN HELCO whose telephone number is (703)756-5539. The examiner can normally be reached on Monday-Friday from 9:00 AM to 5:00 PM.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Matthew Bella, can be reached at telephone number 571-272-7778. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from Patent Center. Status information for published applications may be obtained from Patent Center. Status information for unpublished applications is available through Patent Center for authorized users only. Should you have questions about access to Patent Center, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free).
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) Form at https://www.uspto.gov/patents/uspto-automated- interview-request-air-form.
/NICHOLAS JOHN HELCO/Examiner, Art Unit 2667 /MATTHEW C BELLA/Supervisory Patent Examiner, Art Unit 2667
Prosecution Insights