Patent
US 11,694,307 B2IMAGE ENHANCEMENT SYSTEM AND METHOD BASED ON GENERATIVE ADVERSARIAL NETWORK (GAN) MODEL
Huajie Zeng, Yixiang Fang
Huajie Zeng, Ningbo (CN)·Jul. 4, 2023·US
Copyright © 2026 Matter42. All rights reserved.
Patent
Atlas literature
Patent
US 11,694,307 B2Patent drawings and their descriptions. Click a drawing to enlarge it.
FIG. 1 schematically depicts an image enhancement sys- tem according to an embodiment of the present disclosure;
FIG. 2 schematically depicts an operating state of the image enhancement system according to an embodiment of the present disclosure;
FIG. 3 is a perspective view of a camera of a vehicle with the image enhancement system according to an embodiment of the present disclosure;
FIG. 4 is a perspective view of the camera of the vehicle with the image enhancement system according to an embodiment of the present disclosure from another …
FIG. 5 is a sectional view of the camera of the vehicle with the image enhancement system;
FIG. 6 is a partially enlarged view of
FIG. 7 is an exploded view of the camera of the vehicle with the image enhancement system according to an embodiment of the present disclosure;
FIG. 8 is an exploded view of the camera of the vehicle with the image enhancement system according to an embodiment of the present disclosure from another …
FIG. 9 is a flow chart of the image enhancement system according to an embodiment of the present disclosure.
Claims define the patent's legal scope. Independent claims stand alone; dependent claims (nested) narrow them. Click a claim to expand its dependents.
An image enhancement system based on a generative adversarial network (GAN) model, comprising: an acquiring unit; a training unit; and an enhancement unit; wherein the acquiring unit is configured to acquire a first image of a driving environment captured by a camera of a first vehicle and a second image of the driving environment captured by a camera of a second vehicle; the training unit is configured to acquire a training image based on the first image, wherein the training image comprises a first training image and a second training image, and an image quality of the second training image is lower than that of the first training image; configured to train a discriminative model in the GAN model by using the first training image and the second training image; and configured to train a generative model of the GAN model by using the second training image through a trained discriminative model to obtain an image enhancement model; the enhancement unit is configured to enhance the second image by inputting the second image into the image enhancement model, so as to improve an image quality of the second image; wherein the training unit comprises a processing module, a discriminative model training module and a genera-tive model training module; the processing module is configured to process the first image to obtain the first training image and the second training image; the discriminative model training module is configured to train the discriminative model by using the first training image and the second training image; the generative model training module is configured to train the generative model by using the second training image based on the trained discriminative model to obtain the image enhancement model; wherein the generative model training module comprises an output sub-module, a feature extraction sub-module, a comparison sub-module and a training sub-module; the output sub-module is configured to output an output image after the second training image is input into the generative model; the feature extraction sub-module is configured to extract a feature from feature areas of a desired image, the second training image and the output image; the comparison sub-module is configured to obtain a brightness difference of a specific area between the second training image and the output image, and obtain a texture difference of the specific area between the desired image and the output image; and the training sub-module is configured to train the genera-tive model according to the brightness difference and the texture difference.
The image enhancement system of claim 1, wherein the processing module is configured to classify the first image to obtain the first training image and the second training image.
The image enhancement system of claim 1, wherein the processing module is configured to retain the first image to acquire the first training image and reduce an image quality of the first image to acquire the second training image.
The image enhancement system of claim 1, wherein the specific area is an area corresponding to a road sign.
An image enhancement method based on a GAN model, comprising: (a) acquiring a first image of a driving environment captured by a camera of a first vehicle; (b) acquiring a training image based on the first image, wherein the training image comprises a first training image and a second training image, and an image quality of the second training image is lower than that of the first training image; (c) training a discriminative model in the GAN model using the first training image and the second training image; (d) training a generative model of the GAN model using the second training image based on a trained discrimi-native model to obtain an image enhancement model; (e) acquiring a second image of the driving environment captured by a camera of a second vehicle; and (f) inputting the second image into the image enhance-ment model to improve an image quality of the second image; wherein the step (d) is performed through steps of: (d.1) acquiring a desired image; (d.2) inputting the second training image into the genera-tive model of the GAN model to obtain an output image; (d.3) obtaining a brightness difference of a specific area between the second training image and the output image; (d.4) obtaining a texture difference of the specific area between the desired image and the output image; and (d.5) training the generative model according to the brightness difference and the texture difference.
The image enhancement method of claim 5, wherein in the step (b), the first training image and the second training image are acquired by classifying the first image.
The image enhancement method of claim 5, wherein in the step (b), the first training image is acquired by retaining the first image; and the second training image is acquired by reducing an image quality of the first image.
The image enhancement method of claim 5, wherein the GAN model is trained by a cloud processor to obtain the image enhancement model.
The image enhancement method of claim 5, wherein the specific area is an area corresponding to a road sign. ∗ ∗ ∗ ∗ ∗
Layer stacks claimed or described, ordered top of device to substrate.
image enhancement system based on GAN model
No layer stack recorded.
Measurements and analyses referenced in the patent, with their drawing references.
FIG. 1 schematically depicts an image enhancement sys- tem according to an embodiment of the present disclosure;
Patents and literature cited by this patent (applicant and examiner references).
Cited patents · 5
Related documents with shared materials, methods, properties, or citations.
Patent
Atlas literature
Patent
US 11,694,307 B2Patent drawings and their descriptions. Click a drawing to enlarge it.
FIG. 1 schematically depicts an image enhancement sys- tem according to an embodiment of the present disclosure;
FIG. 2 schematically depicts an operating state of the image enhancement system according to an embodiment of the present disclosure;
FIG. 3 is a perspective view of a camera of a vehicle with the image enhancement system according to an embodiment of the present disclosure;
FIG. 4 is a perspective view of the camera of the vehicle with the image enhancement system according to an embodiment of the present disclosure from another …
FIG. 5 is a sectional view of the camera of the vehicle with the image enhancement system;
FIG. 6 is a partially enlarged view of
FIG. 7 is an exploded view of the camera of the vehicle with the image enhancement system according to an embodiment of the present disclosure;
FIG. 8 is an exploded view of the camera of the vehicle with the image enhancement system according to an embodiment of the present disclosure from another …
FIG. 9 is a flow chart of the image enhancement system according to an embodiment of the present disclosure.
Claims define the patent's legal scope. Independent claims stand alone; dependent claims (nested) narrow them. Click a claim to expand its dependents.
An image enhancement system based on a generative adversarial network (GAN) model, comprising: an acquiring unit; a training unit; and an enhancement unit; wherein the acquiring unit is configured to acquire a first image of a driving environment captured by a camera of a first vehicle and a second image of the driving environment captured by a camera of a second vehicle; the training unit is configured to acquire a training image based on the first image, wherein the training image comprises a first training image and a second training image, and an image quality of the second training image is lower than that of the first training image; configured to train a discriminative model in the GAN model by using the first training image and the second training image; and configured to train a generative model of the GAN model by using the second training image through a trained discriminative model to obtain an image enhancement model; the enhancement unit is configured to enhance the second image by inputting the second image into the image enhancement model, so as to improve an image quality of the second image; wherein the training unit comprises a processing module, a discriminative model training module and a genera-tive model training module; the processing module is configured to process the first image to obtain the first training image and the second training image; the discriminative model training module is configured to train the discriminative model by using the first training image and the second training image; the generative model training module is configured to train the generative model by using the second training image based on the trained discriminative model to obtain the image enhancement model; wherein the generative model training module comprises an output sub-module, a feature extraction sub-module, a comparison sub-module and a training sub-module; the output sub-module is configured to output an output image after the second training image is input into the generative model; the feature extraction sub-module is configured to extract a feature from feature areas of a desired image, the second training image and the output image; the comparison sub-module is configured to obtain a brightness difference of a specific area between the second training image and the output image, and obtain a texture difference of the specific area between the desired image and the output image; and the training sub-module is configured to train the genera-tive model according to the brightness difference and the texture difference.
The image enhancement system of claim 1, wherein the processing module is configured to classify the first image to obtain the first training image and the second training image.
The image enhancement system of claim 1, wherein the processing module is configured to retain the first image to acquire the first training image and reduce an image quality of the first image to acquire the second training image.
The image enhancement system of claim 1, wherein the specific area is an area corresponding to a road sign.
An image enhancement method based on a GAN model, comprising: (a) acquiring a first image of a driving environment captured by a camera of a first vehicle; (b) acquiring a training image based on the first image, wherein the training image comprises a first training image and a second training image, and an image quality of the second training image is lower than that of the first training image; (c) training a discriminative model in the GAN model using the first training image and the second training image; (d) training a generative model of the GAN model using the second training image based on a trained discrimi-native model to obtain an image enhancement model; (e) acquiring a second image of the driving environment captured by a camera of a second vehicle; and (f) inputting the second image into the image enhance-ment model to improve an image quality of the second image; wherein the step (d) is performed through steps of: (d.1) acquiring a desired image; (d.2) inputting the second training image into the genera-tive model of the GAN model to obtain an output image; (d.3) obtaining a brightness difference of a specific area between the second training image and the output image; (d.4) obtaining a texture difference of the specific area between the desired image and the output image; and (d.5) training the generative model according to the brightness difference and the texture difference.
The image enhancement method of claim 5, wherein in the step (b), the first training image and the second training image are acquired by classifying the first image.
The image enhancement method of claim 5, wherein in the step (b), the first training image is acquired by retaining the first image; and the second training image is acquired by reducing an image quality of the first image.
The image enhancement method of claim 5, wherein the GAN model is trained by a cloud processor to obtain the image enhancement model.
The image enhancement method of claim 5, wherein the specific area is an area corresponding to a road sign. ∗ ∗ ∗ ∗ ∗
Layer stacks claimed or described, ordered top of device to substrate.
image enhancement system based on GAN model
No layer stack recorded.
Measurements and analyses referenced in the patent, with their drawing references.
FIG. 1 schematically depicts an image enhancement sys- tem according to an embodiment of the present disclosure;
Patents and literature cited by this patent (applicant and examiner references).
Cited patents · 5
Related documents with shared materials, methods, properties, or citations.
Patent
Atlas literature
Patent
US 11,694,307 B2Patent drawings and their descriptions. Click a drawing to enlarge it.
FIG. 1 schematically depicts an image enhancement sys- tem according to an embodiment of the present disclosure;
FIG. 2 schematically depicts an operating state of the image enhancement system according to an embodiment of the present disclosure;
FIG. 3 is a perspective view of a camera of a vehicle with the image enhancement system according to an embodiment of the present disclosure;
FIG. 4 is a perspective view of the camera of the vehicle with the image enhancement system according to an embodiment of the present disclosure from another …
FIG. 5 is a sectional view of the camera of the vehicle with the image enhancement system;
FIG. 6 is a partially enlarged view of
FIG. 7 is an exploded view of the camera of the vehicle with the image enhancement system according to an embodiment of the present disclosure;
FIG. 8 is an exploded view of the camera of the vehicle with the image enhancement system according to an embodiment of the present disclosure from another …
FIG. 9 is a flow chart of the image enhancement system according to an embodiment of the present disclosure.
Claims define the patent's legal scope. Independent claims stand alone; dependent claims (nested) narrow them. Click a claim to expand its dependents.
An image enhancement system based on a generative adversarial network (GAN) model, comprising: an acquiring unit; a training unit; and an enhancement unit; wherein the acquiring unit is configured to acquire a first image of a driving environment captured by a camera of a first vehicle and a second image of the driving environment captured by a camera of a second vehicle; the training unit is configured to acquire a training image based on the first image, wherein the training image comprises a first training image and a second training image, and an image quality of the second training image is lower than that of the first training image; configured to train a discriminative model in the GAN model by using the first training image and the second training image; and configured to train a generative model of the GAN model by using the second training image through a trained discriminative model to obtain an image enhancement model; the enhancement unit is configured to enhance the second image by inputting the second image into the image enhancement model, so as to improve an image quality of the second image; wherein the training unit comprises a processing module, a discriminative model training module and a genera-tive model training module; the processing module is configured to process the first image to obtain the first training image and the second training image; the discriminative model training module is configured to train the discriminative model by using the first training image and the second training image; the generative model training module is configured to train the generative model by using the second training image based on the trained discriminative model to obtain the image enhancement model; wherein the generative model training module comprises an output sub-module, a feature extraction sub-module, a comparison sub-module and a training sub-module; the output sub-module is configured to output an output image after the second training image is input into the generative model; the feature extraction sub-module is configured to extract a feature from feature areas of a desired image, the second training image and the output image; the comparison sub-module is configured to obtain a brightness difference of a specific area between the second training image and the output image, and obtain a texture difference of the specific area between the desired image and the output image; and the training sub-module is configured to train the genera-tive model according to the brightness difference and the texture difference.
The image enhancement system of claim 1, wherein the processing module is configured to classify the first image to obtain the first training image and the second training image.
The image enhancement system of claim 1, wherein the processing module is configured to retain the first image to acquire the first training image and reduce an image quality of the first image to acquire the second training image.
The image enhancement system of claim 1, wherein the specific area is an area corresponding to a road sign.
An image enhancement method based on a GAN model, comprising: (a) acquiring a first image of a driving environment captured by a camera of a first vehicle; (b) acquiring a training image based on the first image, wherein the training image comprises a first training image and a second training image, and an image quality of the second training image is lower than that of the first training image; (c) training a discriminative model in the GAN model using the first training image and the second training image; (d) training a generative model of the GAN model using the second training image based on a trained discrimi-native model to obtain an image enhancement model; (e) acquiring a second image of the driving environment captured by a camera of a second vehicle; and (f) inputting the second image into the image enhance-ment model to improve an image quality of the second image; wherein the step (d) is performed through steps of: (d.1) acquiring a desired image; (d.2) inputting the second training image into the genera-tive model of the GAN model to obtain an output image; (d.3) obtaining a brightness difference of a specific area between the second training image and the output image; (d.4) obtaining a texture difference of the specific area between the desired image and the output image; and (d.5) training the generative model according to the brightness difference and the texture difference.
The image enhancement method of claim 5, wherein in the step (b), the first training image and the second training image are acquired by classifying the first image.
The image enhancement method of claim 5, wherein in the step (b), the first training image is acquired by retaining the first image; and the second training image is acquired by reducing an image quality of the first image.
The image enhancement method of claim 5, wherein the GAN model is trained by a cloud processor to obtain the image enhancement model.
The image enhancement method of claim 5, wherein the specific area is an area corresponding to a road sign. ∗ ∗ ∗ ∗ ∗
Layer stacks claimed or described, ordered top of device to substrate.
image enhancement system based on GAN model
No layer stack recorded.
Measurements and analyses referenced in the patent, with their drawing references.
FIG. 1 schematically depicts an image enhancement sys- tem according to an embodiment of the present disclosure;
Patents and literature cited by this patent (applicant and examiner references).
Cited patents · 5
Related documents with shared materials, methods, properties, or citations.
Patent
Atlas literature
Patent
US 11,694,307 B2Patent drawings and their descriptions. Click a drawing to enlarge it.
FIG. 1 schematically depicts an image enhancement sys- tem according to an embodiment of the present disclosure;
FIG. 2 schematically depicts an operating state of the image enhancement system according to an embodiment of the present disclosure;
FIG. 3 is a perspective view of a camera of a vehicle with the image enhancement system according to an embodiment of the present disclosure;
FIG. 4 is a perspective view of the camera of the vehicle with the image enhancement system according to an embodiment of the present disclosure from another …
FIG. 5 is a sectional view of the camera of the vehicle with the image enhancement system;
FIG. 6 is a partially enlarged view of
FIG. 7 is an exploded view of the camera of the vehicle with the image enhancement system according to an embodiment of the present disclosure;
FIG. 8 is an exploded view of the camera of the vehicle with the image enhancement system according to an embodiment of the present disclosure from another …
FIG. 9 is a flow chart of the image enhancement system according to an embodiment of the present disclosure.
Claims define the patent's legal scope. Independent claims stand alone; dependent claims (nested) narrow them. Click a claim to expand its dependents.
An image enhancement system based on a generative adversarial network (GAN) model, comprising: an acquiring unit; a training unit; and an enhancement unit; wherein the acquiring unit is configured to acquire a first image of a driving environment captured by a camera of a first vehicle and a second image of the driving environment captured by a camera of a second vehicle; the training unit is configured to acquire a training image based on the first image, wherein the training image comprises a first training image and a second training image, and an image quality of the second training image is lower than that of the first training image; configured to train a discriminative model in the GAN model by using the first training image and the second training image; and configured to train a generative model of the GAN model by using the second training image through a trained discriminative model to obtain an image enhancement model; the enhancement unit is configured to enhance the second image by inputting the second image into the image enhancement model, so as to improve an image quality of the second image; wherein the training unit comprises a processing module, a discriminative model training module and a genera-tive model training module; the processing module is configured to process the first image to obtain the first training image and the second training image; the discriminative model training module is configured to train the discriminative model by using the first training image and the second training image; the generative model training module is configured to train the generative model by using the second training image based on the trained discriminative model to obtain the image enhancement model; wherein the generative model training module comprises an output sub-module, a feature extraction sub-module, a comparison sub-module and a training sub-module; the output sub-module is configured to output an output image after the second training image is input into the generative model; the feature extraction sub-module is configured to extract a feature from feature areas of a desired image, the second training image and the output image; the comparison sub-module is configured to obtain a brightness difference of a specific area between the second training image and the output image, and obtain a texture difference of the specific area between the desired image and the output image; and the training sub-module is configured to train the genera-tive model according to the brightness difference and the texture difference.
The image enhancement system of claim 1, wherein the processing module is configured to classify the first image to obtain the first training image and the second training image.
The image enhancement system of claim 1, wherein the processing module is configured to retain the first image to acquire the first training image and reduce an image quality of the first image to acquire the second training image.
The image enhancement system of claim 1, wherein the specific area is an area corresponding to a road sign.
An image enhancement method based on a GAN model, comprising: (a) acquiring a first image of a driving environment captured by a camera of a first vehicle; (b) acquiring a training image based on the first image, wherein the training image comprises a first training image and a second training image, and an image quality of the second training image is lower than that of the first training image; (c) training a discriminative model in the GAN model using the first training image and the second training image; (d) training a generative model of the GAN model using the second training image based on a trained discrimi-native model to obtain an image enhancement model; (e) acquiring a second image of the driving environment captured by a camera of a second vehicle; and (f) inputting the second image into the image enhance-ment model to improve an image quality of the second image; wherein the step (d) is performed through steps of: (d.1) acquiring a desired image; (d.2) inputting the second training image into the genera-tive model of the GAN model to obtain an output image; (d.3) obtaining a brightness difference of a specific area between the second training image and the output image; (d.4) obtaining a texture difference of the specific area between the desired image and the output image; and (d.5) training the generative model according to the brightness difference and the texture difference.
The image enhancement method of claim 5, wherein in the step (b), the first training image and the second training image are acquired by classifying the first image.
The image enhancement method of claim 5, wherein in the step (b), the first training image is acquired by retaining the first image; and the second training image is acquired by reducing an image quality of the first image.
The image enhancement method of claim 5, wherein the GAN model is trained by a cloud processor to obtain the image enhancement model.
The image enhancement method of claim 5, wherein the specific area is an area corresponding to a road sign. ∗ ∗ ∗ ∗ ∗
Layer stacks claimed or described, ordered top of device to substrate.
image enhancement system based on GAN model
No layer stack recorded.
Measurements and analyses referenced in the patent, with their drawing references.
FIG. 1 schematically depicts an image enhancement sys- tem according to an embodiment of the present disclosure;
Patents and literature cited by this patent (applicant and examiner references).
Cited patents · 5
Related documents with shared materials, methods, properties, or citations.