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 .
Response to Remarks
The Office Action has been made issued in response to amendment filed July 16, 2026. Claims 1-3, 5-8 and 10 are pending. Applicant’s arguments have been carefully and respectfully considered in light of the instant amendment, and are not persuasive. Accordingly, this action has been made FINAL.
35 USC § 103 Claim Rejection
On pages 6-7 of the Response, Applicant argues that Guan does not disclose or suggests decomposition level, filter type selection, or processing intensity as recited by claim 1. Further, Applicant argue that “Guan's two sub-networks are a fixed structural feature of the NODE architecture, present in the same configuration regardless of the content of the input image, rather than parameters that are dynamically adjusted based on analyzed image characteristics. Guan's sub- networks also operate to estimate different categories of noise as part of the denoising stage; they do not decompose the raw input image into frequency components, and Guan is silent regarding decomposition level, filter type, or processing intensity”. The Examiner disagrees with Applicant.
Guan teaches decompose the noise into Gaussian + Poisson noise and defective pixel noise, using dedicated sub-networks trained in a multi-task setting (see [p][002]), Guan also teaches that stochastic noise can be modelled with a Gaussian distribution (see [p][)) which is equivalent to a filter type selection. Moreover, Guan teaches the brightness dependent (see section 1, [p][002) which is dynamic adjustment. Finally, Guan teaches wherein the decomposition adaptively selects processing parameters based on the image characteristics in section 2.3, [p][005]) which teaches that “our designed neural network can simultaneously estimate Gaussian+Poisson noise and defective pixel noise using two separate sub-networks. This way, different parts of the network focus on specific types of noise. To our knowledge, the defective pixel noise removal using a deep neural network has never done before”.
Thus, Guan teaches preprocessing parameters based on the image characteristics, wherein the determined preprocessing parameters include at least one parameter selected from decomposition level, filter type selection, and processing intensity, and wherein the parameters are dynamically adjusted based on the image characteristics.
Applicant arguments with respect to claims 6-10 are similar to those for claim 1 -5 and the argument applies. +
Terminal Disclaimer
The terminal disclaimer filed on July 16, 2026 disclaiming the terminal portion of any patent granted on this application which would extend beyond the expiration date of US Patent No.: 12,175,638 and US Patent No.:12394018 has been reviewed and is accepted. The terminal disclaimer has been recorded.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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 of this title, 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 1-5 are rejected under 35 U.S.C. 103 as being unpatentable over Mengchuan et al (English Translation of CN 115578281 A) in view of Guan et al (NPL titled: NODE: Extreme Low Light Raw Image Denoising using a Noise Decomposition Network) in view of Moran et al (Pub No.: 20220036523).
Regarding independent claim 1, Mengchuan teaches a computer system for low-light image enhancement ([m]ethod for processing original image by using neural network based on discrete wavelet transform by processing high-quality images which are comparable to those shot by a DSLR camera can be regenerated from low-quality original files of the smart phone sensor – see abstract and page 2, section Background, [p][003]), comprising:
receive a raw input image (a bayer RAW image – see page 7, 2nd full para) captured under low-light conditions (low-quality original files of the smart phone sensor – see page 2, section Background, [p][003]);
decompose the raw input image into a plurality of frequency components using the determined preprocessing parameters (separating an image low-frequency part and an image high-frequency part from an input image by using discrete wavelet transform – see page 6, [p][001], step 1),
process each frequency component using a machine learning model trained for denoising to generate enhanced frequency components (processing the high-frequency part of the image by using a residual error network I with N1 residual error blocks and M1 channels in each residual error block; for the low-frequency part of the image, processing by using a residual error network II with N2 residual error blocks and M2 channels in each residual error block, wherein N2 is more than N1, and M2 is more than M1 – see page 6, [p][001]);
reconstruct an enhanced image from the enhanced frequency components (synthesizing the data information output by the first residual error network and the second residual error network into an output image by adopting inverse discrete wavelet transform – see page 6, 1st para);
and provide the enhanced image to an image processing pipeline (characterized in that an end-to-end RAW to RGB convolutional neural network is adopted to replace the whole image processing pipeline of a smartphone camera module to process an input image – see page 6, [p][001]).
Mengchuan does not explicitly teach analyze the raw input image to determine one or more image characteristics selected from brightness levels, contrast levels, noise estimation, and detail complexity; determine preprocessing parameters based on the determined image characteristics, wherein the determined preprocessing parameters include at least one parameter selected from decomposition level, filter type selection, and processing intensity, and wherein the parameters are dynamically adjusted based on the image characteristic and wherein the decomposition adaptively selects processing parameters based on the image characteristics.
However, Guan explicitly teaches analyze the raw input image to determine one or more image characteristics selected from brightness levels, contrast levels, noise estimation ([t]here are several types of noise present in raw data. Read noise results from thermal and electrical noise in the electronics of the imaging sensor. This stochastic noise can be modelled with a Gaussian distribution. Shot noise is related to the number of photons arriving at the sensor, and is therefore brightness dependent. Shot noise can be effectively modeled with a Poisson distribution. Defective pixel noise arises from different sensitivities of pixels on the sensor to incoming light – see section 1, [p][002]), and detail complexity; determine preprocessing parameters based on the determined image characteristics (decompose the noise into Gaussian + Poisson noise and defective pixel noise, using dedicated sub-networks trained in a multi-task setting – see [p][002]), wherein the determined preprocessing parameters include at least one parameter selected from decomposition level, filter type selection, and processing intensity, and wherein the parameters are dynamically adjusted based on the image characteristic (This stochastic noise can be modelled with a Gaussian distribution. Shot noise is related to the number of photons arriving at the sensor, and is therefore brightness dependent. Shot noise can be effectively modeled with a Poisson distribution. Defective pixel noise arises from different sensitivities of pixels on the sensor to incoming light – see section 1, [p][002) and wherein the decomposition adaptively selects processing parameters based on the image characteristics ([o]ur designed neural network can simultaneously estimate Gaussian+Poisson noise and defective pixel noise using two separate sub-networks. This way, different parts of the network focus on specific types of noise. To our knowledge, the defective pixel noise removal using a deep neural network has never done before – see section 2.3, [p][005]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Mengchuan of computer system for low-light image enhancement, comprising: a hardware memory, with the teachings of Guan analyze the raw input image to determine one or more image characteristics selected from brightness levels, contrast levels, noise estimation, and detail complexity; determine preprocessing parameters based on the determined image characteristics and wherein the decomposition adaptively selects processing parameters based on the image characteristics
Wherein having Mengchuan analyze the raw input image to determine one or more image characteristics selected from brightness levels, contrast levels, noise estimation, and detail complexity; determine preprocessing parameters based on the determined image characteristics and wherein the decomposition adaptively selects processing parameters based on the image characteristics.
. The motivation behind the modification would have been for implementing an end-to-end RAW to RGB convolutional neural network by separately estimates defective pixel noise, in conjunction with Gaussian and Poisson noise, to denoise an extreme low light image since both Mengchuan and Guan relates to image denoising, wherein Mengchuan for processing an original image by a neural network based on discrete wavelet transform, which is characterized in that an end-to-end RAW to RGB convolutional neural network is adopted to replace the whole image processing pipeline of a smartphone camera module to process an input image while Guan performs a multi-task deep neural network called Noise Decomposition (NODE) that explicitly and separately estimates defective pixel noise, in conjunction with Gaussian and Poisson noise, to denoise an extreme low light image (Please see Mengchuan et al (English Translation of CN 115578281 A) see page 6, [p][001] and Guan et al (NPL titled: NODE: Extreme Low Light Raw Image Denoising using a Noise Decomposition Network), see Abstract).
Mengchuan in view of Guan does not explicitly teach a hardware memory, wherein the computer system is configured to execute software instructions on nontransitory machine-readable storage media that:
Moran explicitly teaches a hardware memory (memory – see [p][0077]), wherein the computer system is configured to execute software instructions (memory stores in a non-transient way code – see [p][0077]) on nontransitory machine-readable storage media that (see [p][077]):
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Mengchuan in view of Guan of computer system for low-light image enhancement, comprising: a hardware memory, with the teachings of Moran a hardware memory, wherein the computer system is configured to execute software instructions on nontransitory machine-readable storage media that:
Wherein having Guan a hardware memory, wherein the computer system is configured to execute software instructions on nontransitory machine-readable storage media that:.
. The motivation behind the modification would have been for implementing an end-to-end RAW to RGB convolutional neural network by separately estimates defective pixel noise, in conjunction with Gaussian and Poisson noise, to improve image quality by allowing for independent adjustment of properties in different color spaces of the image since both Mengchuan and Guan relates to image denoising, wherein Mengchuan for processing an original image by a neural network based on discrete wavelet transform, which is characterized in that an end-to-end RAW to RGB convolutional neural network is adopted to replace the whole image processing pipeline of a smartphone camera module to process an input image while Moran to improve image quality by allowing for independent adjustment of properties in different color spaces of the image (Please see Mengchuan et al (English Translation of CN 115578281 A) see page 6, [p][001] and Moran et al (Pub No.: US20220036523A1), see [p][0010]).
Regarding claim 2, Mengchuan in view of Guan and Moran teach the computer system of claim 1; Mengchuan in view of Moran does not explicitly teach wherein decomposing the raw input image comprises performing a wavelet decomposition process that selects from multiple wavelet types based on the determined preprocessing parameters.
Guan explicitly teaches wherein decomposing the raw input image comprises performing a wavelet decomposition process that selects from multiple wavelet types based on the determined preprocessing parameters ([o]ur designed neural network can simultaneously estimate Gaussian+Poisson noise and defective pixel noise using two separate sub-networks. This way, different parts of the network focus on specific types of noise. To our knowledge, the defective pixel noise removal using a deep neural network has never done before – see section 2.3, [p][005]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Mengchuan as modified by Moran of computer system for low-light image enhancement, comprising: a hardware memory, with the teachings of Guan wherein decomposing the raw input image comprises performing a wavelet decomposition process that selects from multiple wavelet types based on the determined preprocessing parameters
Wherein having Mengchuan wherein decomposing the raw input image comprises performing a wavelet decomposition process that selects from multiple wavelet types based on the determined preprocessing parameters.
. The motivation behind the modification would have been for implementing an end-to-end RAW to RGB convolutional neural network by separately estimates defective pixel noise, in conjunction with Gaussian and Poisson noise, to denoise an extreme low light image since both Mengchuan and Guan relates to image denoising, wherein Mengchuan for processing an original image by a neural network based on discrete wavelet transform, which is characterized in that an end-to-end RAW to RGB convolutional neural network is adopted to replace the whole image processing pipeline of a smartphone camera module to process an input image while Guan performs a multi-task deep neural network called Noise Decomposition (NODE) that explicitly and separately estimates defective pixel noise, in conjunction with Gaussian and Poisson noise, to denoise an extreme low light image (Please see Mengchuan et al (English Translation of CN 115578281 A) see page 6, [p][001] and Guan et al (NPL titled: NODE: Extreme Low Light Raw Image Denoising using a Noise Decomposition Network), see Abstract).
Regarding claim 3, Mengchuan in view of Guan and Moran teach the computer system of claim 1, Mengchuan teaches wherein the raw input image comprises a Bayer format image (a bayer RAW image – see page 7, 2nd full para), and the computer system creates subsampled subimages from the Bayer format image (a bayer RAW image… processed into a four-channel RGGB image, see page 7, 2nd full para).
Regarding claim 4, Mengchuan in view of Guan and Moran teach the computer system of claim 1, Mengchuan in view of Moran does not explicitly teach wherein the determined preprocessing parameters include at least one parameter selected from decomposition level, filter type selection, and processing intensity, and wherein the parameters are dynamically adjusted based on the image characteristics.
However, Guan explicitly teaches wherein the determined preprocessing parameters include at least one parameter selected from decomposition level, filter type selection, and processing intensity, and wherein the parameters are dynamically adjusted based on the image characteristic ([t]here are several types of noise present in raw data. Read noise results from thermal and electrical noise in the electronics of the imaging sensor. This stochastic noise can be modelled with a Gaussian distribution. Shot noise is related to the number of photons arriving at the sensor, and is therefore brightness dependent. Shot noise can be effectively modeled with a Poisson distribution. Defective pixel noise arises from different sensitivities of pixels on the sensor to incoming light – see section 1, [p][002] and decompose the noise into Gaussian + Poisson noise and defective pixel noise, using dedicated sub-networks trained in a multi-task setting – see [p][002]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Mengchuan in view of Moran of computer system for low-light image enhancement, comprising: a hardware memory, with the teachings of Guan wherein the determined preprocessing parameters include at least one parameter selected from decomposition level, filter type selection, and processing intensity, and wherein the parameters are dynamically adjusted based on the image characteristic
Wherein having Mengchuan wherein the determined preprocessing parameters include at least one parameter selected from decomposition level, filter type selection, and processing intensity, and wherein the parameters are dynamically adjusted based on the image characteristic.
. The motivation behind the modification would have been for implementing an end-to-end RAW to RGB convolutional neural network by separately estimates defective pixel noise, in conjunction with Gaussian and Poisson noise, to denoise an extreme low light image since both Mengchuan and Guan relates to image denoising, wherein Mengchuan for processing an original image by a neural network based on discrete wavelet transform, which is characterized in that an end-to-end RAW to RGB convolutional neural network is adopted to replace the whole image processing pipeline of a smartphone camera module to process an input image while Guan performs a multi-task deep neural network called Noise Decomposition (NODE) that explicitly and separately estimates defective pixel noise, in conjunction with Gaussian and Poisson noise, to denoise an extreme low light image (Please see Mengchuan et al (English Translation of CN 115578281 A) see page 6, [p][001] and Guan et al (NPL titled: NODE: Extreme Low Light Raw Image Denoising using a Noise Decomposition Network), see Abstract).
Regarding claim 5, Mengchuan in view of Guan and Moran teach the computer system of claim 1, Mengchuan teaches wherein the machine learning model comprises neural networks, each neural network including at least one activation function selected from Leaky ReLU and ReLU activation functions (see page 7, 3rd full para).
Claims 6-10 are rejected under 35 U.S.C. 103 as being unpatentable over Mengchuan et al (English Translation of CN 115578281 A) in view of Guan et al (NPL titled: NODE: Extreme Low Light Raw Image Denoising using a Noise Decomposition Network).
Regarding independent claim 6, Mengchuan teaches computer-implemented method for enhancing low-light images ([m]ethod for processing original image by using neural network based on discrete wavelet transform by processing high-quality images which are comparable to those shot by a DSLR camera can be regenerated from low-quality original files of the smart phone sensor – see abstract and page 2, section Background, [p][003]), comprising:
receive a raw input image (a bayer RAW image – see page 7, 2nd full para) captured under low-light conditions (low-quality original files of the smart phone sensor – see page 2, section Background, [p][003]);
decompose the raw input image into a plurality of frequency components using the determined preprocessing parameters (separating an image low-frequency part and an image high-frequency part from an input image by using discrete wavelet transform – see page 6, [p][001], step 1),
process each frequency component using a machine learning model trained for denoising to generate enhanced frequency components (processing the high-frequency part of the image by using a residual error network I with N1 residual error blocks and M1 channels in each residual error block; for the low-frequency part of the image, processing by using a residual error network II with N2 residual error blocks and M2 channels in each residual error block, wherein N2 is more than N1, and M2 is more than M1 – see page 6, [p][001]);
and reconstruct an enhanced image from the enhanced frequency components (synthesizing the data information output by the first residual error network and the second residual error network into an output image by adopting inverse discrete wavelet transform – see page 6, 1st para).
Mengchuan does not explicitly teach analyze the raw input image to determine one or more image characteristics selected from brightness levels, contrast levels, noise estimation, and detail complexity; determine preprocessing parameters based on the determined image characteristics, wherein the determined preprocessing parameters include at least one parameter selected from decomposition level, filter type selection, and processing intensity, and wherein the parameters are dynamically adjusted based on the image characteristic and wherein the decomposition adaptively selects processing parameters based on the image characteristics.
However, Guan explicitly teaches analyze the raw input image to determine one or more image characteristics selected from brightness levels, contrast levels, noise estimation ([t]here are several types of noise present in raw data. Read noise results from thermal and electrical noise in the electronics of the imaging sensor. This stochastic noise can be modelled with a Gaussian distribution. Shot noise is related to the number of photons arriving at the sensor, and is therefore brightness dependent. Shot noise can be effectively modeled with a Poisson distribution. Defective pixel noise arises from different sensitivities of pixels on the sensor to incoming light – see section 1, [p][002]), and detail complexity; determine preprocessing parameters based on the determined image characteristics (decompose the noise into Gaussian + Poisson noise and defective pixel noise, using dedicated sub-networks trained in a multi-task setting – see [p][002]), wherein the determined preprocessing parameters include at least one parameter selected from decomposition level, filter type selection, and processing intensity, and wherein the parameters are dynamically adjusted based on the image characteristic (This stochastic noise can be modelled with a Gaussian distribution. Shot noise is related to the number of photons arriving at the sensor, and is therefore brightness dependent. Shot noise can be effectively modeled with a Poisson distribution. Defective pixel noise arises from different sensitivities of pixels on the sensor to incoming light – see section 1, [p][002) and wherein the decomposition adaptively selects processing parameters based on the image characteristics ([o]ur designed neural network can simultaneously estimate Gaussian+Poisson noise and defective pixel noise using two separate sub-networks. This way, different parts of the network focus on specific types of noise. To our knowledge, the defective pixel noise removal using a deep neural network has never done before – see section 2.3, [p][005]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Mengchuan of computer system for low-light image enhancement, comprising: a hardware memory, with the teachings of Guan analyze the raw input image to determine one or more image characteristics selected from brightness levels, contrast levels, noise estimation, and detail complexity; determine preprocessing parameters based on the determined image characteristics and wherein the decomposition adaptively selects processing parameters based on the image characteristics
Wherein having Mengchuan analyze the raw input image to determine one or more image characteristics selected from brightness levels, contrast levels, noise estimation, and detail complexity; determine preprocessing parameters based on the determined image characteristics and wherein the decomposition adaptively selects processing parameters based on the image characteristics.
. The motivation behind the modification would have been for implementing an end-to-end RAW to RGB convolutional neural network by separately estimates defective pixel noise, in conjunction with Gaussian and Poisson noise, to denoise an extreme low light image since both Mengchuan and Guan relates to image denoising, wherein Mengchuan for processing an original image by a neural network based on discrete wavelet transform, which is characterized in that an end-to-end RAW to RGB convolutional neural network is adopted to replace the whole image processing pipeline of a smartphone camera module to process an input image while Guan performs a multi-task deep neural network called Noise Decomposition (NODE) that explicitly and separately estimates defective pixel noise, in conjunction with Gaussian and Poisson noise, to denoise an extreme low light image (Please see Mengchuan et al (English Translation of CN 115578281 A) see page 6, [p][001] and Guan et al (NPL titled: NODE: Extreme Low Light Raw Image Denoising using a Noise Decomposition Network), see Abstract).
Regarding claim 7, Mengchuan in view of Guan teach the computer-implemented method of claim 6; Mengchuan does not explicitly teach wherein decomposing the raw input image comprises performing a wavelet decomposition process that selects from multiple wavelet types based on the determined preprocessing parameters.
Guan explicitly teaches wherein decomposing the raw input image comprises performing a wavelet decomposition process that selects from multiple wavelet types based on the determined preprocessing parameters ([o]ur designed neural network can simultaneously estimate Gaussian+Poisson noise and defective pixel noise using two separate sub-networks. This way, different parts of the network focus on specific types of noise. To our knowledge, the defective pixel noise removal using a deep neural network has never done before – see section 2.3, [p][005]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Mengchuan of computer system for low-light image enhancement, comprising: a hardware memory, with the teachings of Guan wherein decomposing the raw input image comprises performing a wavelet decomposition process that selects from multiple wavelet types based on the determined preprocessing parameters
Wherein having Mengchuan wherein decomposing the raw input image comprises performing a wavelet decomposition process that selects from multiple wavelet types based on the determined preprocessing parameters.
. The motivation behind the modification would have been for implementing an end-to-end RAW to RGB convolutional neural network by separately estimates defective pixel noise, in conjunction with Gaussian and Poisson noise, to denoise an extreme low light image since both Mengchuan and Guan relates to image denoising, wherein Mengchuan for processing an original image by a neural network based on discrete wavelet transform, which is characterized in that an end-to-end RAW to RGB convolutional neural network is adopted to replace the whole image processing pipeline of a smartphone camera module to process an input image while Guan performs a multi-task deep neural network called Noise Decomposition (NODE) that explicitly and separately estimates defective pixel noise, in conjunction with Gaussian and Poisson noise, to denoise an extreme low light image (Please see Mengchuan et al (English Translation of CN 115578281 A) see page 6, [p][001] and Guan et al (NPL titled: NODE: Extreme Low Light Raw Image Denoising using a Noise Decomposition Network), see Abstract).
Regarding claim 8, Mengchuan in view of Guan teach the computer-implemented method of claim 6, Mengchuan teaches wherein the raw input image comprises a Bayer format image (a bayer RAW image – see page 7, 2nd full para), and the computer system creates subsampled subimages from the Bayer format image (a bayer RAW image… processed into a four-channel RGGB image, see page 7, 2nd full para).
Regarding claim 9, Mengchuan in view of Guan teach the computer-implemented method of claim 6, Mengchuan does not explicitly teach wherein the determined preprocessing parameters include at least one parameter selected from decomposition level, filter type selection, and processing intensity, and wherein the parameters are dynamically adjusted based on the image characteristics.
However, Guan explicitly teaches wherein the determined preprocessing parameters include at least one parameter selected from decomposition level, filter type selection, and processing intensity, and wherein the parameters are dynamically adjusted based on the image characteristic ([t]here are several types of noise present in raw data. Read noise results from thermal and electrical noise in the electronics of the imaging sensor. This stochastic noise can be modelled with a Gaussian distribution. Shot noise is related to the number of photons arriving at the sensor, and is therefore brightness dependent. Shot noise can be effectively modeled with a Poisson distribution. Defective pixel noise arises from different sensitivities of pixels on the sensor to incoming light – see section 1, [p][002] and decompose the noise into Gaussian + Poisson noise and defective pixel noise, using dedicated sub-networks trained in a multi-task setting – see [p][002]).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of Mengchuan of computer system for low-light image enhancement, comprising: a hardware memory, with the teachings of Guan wherein the determined preprocessing parameters include at least one parameter selected from decomposition level, filter type selection, and processing intensity, and wherein the parameters are dynamically adjusted based on the image characteristic
Wherein having Mengchuan wherein the determined preprocessing parameters include at least one parameter selected from decomposition level, filter type selection, and processing intensity, and wherein the parameters are dynamically adjusted based on the image characteristic.
. The motivation behind the modification would have been for implementing an end-to-end RAW to RGB convolutional neural network by separately estimates defective pixel noise, in conjunction with Gaussian and Poisson noise, to denoise an extreme low light image since both Mengchuan and Guan relates to image denoising, wherein Mengchuan for processing an original image by a neural network based on discrete wavelet transform, which is characterized in that an end-to-end RAW to RGB convolutional neural network is adopted to replace the whole image processing pipeline of a smartphone camera module to process an input image while Guan performs a multi-task deep neural network called Noise Decomposition (NODE) that explicitly and separately estimates defective pixel noise, in conjunction with Gaussian and Poisson noise, to denoise an extreme low light image (Please see Mengchuan et al (English Translation of CN 115578281 A) see page 6, [p][001] and Guan et al (NPL titled: NODE: Extreme Low Light Raw Image Denoising using a Noise Decomposition Network), see Abstract).
Regarding claim 10, Mengchuan in view of Guan teach the computer-implemented method of claim 6, Mengchuan teaches wherein the machine learning model comprises neural networks, each neural network including at least one activation function selected from Leaky ReLU and ReLU activation functions (see page 7, 3rd full para).
Conclusion
THIS ACTION IS MADE FINAL. 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.
Inquiries
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ANDRAE S ALLISON whose telephone number is (571)270-1052. The examiner can normally be reached on Monday-Friday 9am-5pm EST.
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) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Chineyere Wills-Burns, can be reached on (571) 272-9752. 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 the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/ANDRAE S ALLISON/Primary Examiner, Art Unit 2673
August 9, 2026
Prosecution Insights