CN116385318A - Image quality enhancement method and system based on cloud desktop - Google Patents
Image quality enhancement method and system based on cloud desktop Download PDFInfo
- Publication number
- CN116385318A CN116385318A CN202310661815.3A CN202310661815A CN116385318A CN 116385318 A CN116385318 A CN 116385318A CN 202310661815 A CN202310661815 A CN 202310661815A CN 116385318 A CN116385318 A CN 116385318A
- Authority
- CN
- China
- Prior art keywords
- image
- cloud desktop
- super
- desktop end
- resolution reconstruction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 230000006870 function Effects 0.000 claims description 18
- 230000004044 response Effects 0.000 claims description 18
- 238000001914 filtration Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 4
- 230000003044 adaptive effect Effects 0.000 claims description 3
- 239000002131 composite material Substances 0.000 claims description 3
- 238000003708 edge detection Methods 0.000 claims description 3
- 230000005484 gravity Effects 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 claims 2
- 230000005540 biological transmission Effects 0.000 abstract description 10
- 238000012545 processing Methods 0.000 abstract description 10
- 230000000007 visual effect Effects 0.000 abstract description 9
- 230000006872 improvement Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 230000006835 compression Effects 0.000 description 4
- 238000007906 compression Methods 0.000 description 4
- 230000002708 enhancing effect Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4053—Super resolution, i.e. output image resolution higher than sensor resolution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration by the use of local operators
-
- G06T5/90—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a cloud desktop-based image quality enhancement method and a cloud desktop-based image quality enhancement system, which comprise the following steps: s1: acquiring an original image and a cloud desktop end image transmitted to a cloud desktop, and denoising the image to obtain the denoised original image and the cloud desktop end image; s2: extracting image gradients of the denoised original image and the cloud desktop end image; s3: constructing a super-resolution reconstruction network based on the original image after denoising in the step S1, the cloud desktop end image and the image extracted in the step S2; s4: training the network constructed in the step S3, and obtaining a cloud desktop end image with improved definition based on the super-resolution reconstruction network after training; s5: and carrying out dynamic range adjustment on the cloud desktop end image with improved definition to obtain the cloud desktop end image with enhanced final image quality. The invention has the advantages of improving the definition of the image, improving the visual experience, improving the usability of the application, maintaining the transmission and processing efficiency, providing the self-adaptability and the customizability, and the like.
Description
Technical Field
The invention belongs to the field of image quality enhancement, and particularly relates to an image quality enhancement method and system based on a cloud desktop.
Background
In a cloud desktop environment, image quality is critical to the user experience and requirements of the application scene. However, due to limitations in transmission and processing, the cloud desktop end image may be affected by problems such as noise, low definition, and dynamic range limitations, reducing the quality and look and feel of the image. Therefore, a method for improving the image quality of the cloud desktop end image is needed, and the definition improvement and the visual effect enhancement of the image are realized through technologies such as denoising, super-resolution reconstruction and dynamic range adjustment. In the prior art, some image enhancement methods already exist, but they may have some limitations such as low processing efficiency, high complexity, image quality loss, and the like. The methods involve complex algorithms and calculations in the image enhancement process, resulting in long processing times and inapplicability to cloud desktop environments with high real-time requirements. Meanwhile, these methods may introduce additional noise or artifacts while improving the definition of the image, resulting in loss of image quality or unnatural vision.
Disclosure of Invention
In view of the above, the invention provides a method and a system for enhancing image quality based on a cloud desktop, which aim to provide higher quality visual experience for cloud desktop end images on the premise of ensuring transmission and processing efficiency through the steps of denoising, super-resolution reconstruction, dynamic range adjustment and the like.
The image quality enhancement method based on the cloud desktop provided by the invention comprises the following steps of:
s1: acquiring an original image and a cloud desktop end image transmitted to a cloud desktop, and denoising the image to obtain the denoised original image and the cloud desktop end image;
s2: extracting image gradients of the denoised original image and the cloud desktop end image;
s3: constructing a super-resolution reconstruction network based on the original image after denoising in the step S1, the cloud desktop end image and the image extracted in the step S2;
s4: training the network constructed in the step S3, and obtaining a cloud desktop end image with improved definition based on the super-resolution reconstruction network after training;
s5: and carrying out dynamic range adjustment on the cloud desktop end image with improved definition to obtain the cloud desktop end image with enhanced final image quality.
As a further improvement of the present invention:
optionally, the step S1 obtains an original image and a cloud desktop end image transmitted to a cloud desktop, and denoises the image to obtain the denoised original image and the cloud desktop end image, including:
the method comprises the steps of obtaining an original image and a cloud desktop end image transmitted to a cloud desktop, wherein the obtained image data are as follows:
wherein ,
and />
Respectively obtaining an ith original image and a cloud desktop end image;
denoising an original image and a cloud desktop end image based on self-adaptive space-time median filtering, wherein the self-adaptive space-time median filtering comprises the following steps:
s11: calculating the adaptive filter window size:
wherein, size represents the size of the filtering window, th is the window screening threshold, and the calculation mode is as follows:
wherein ,
，
，
respectively the maximum value, the minimum value and the average value of the ith image;
is the mean value of the i-1 th image;
s12: calculating a space-time median filtering result:
order the
For the median filter window of the ith image,
for the median filter window of the i-1 th image, the median filter results are:
wherein ,
calculating the median value of the input sequence by the function;
，
and
the method comprises the steps of respectively obtaining an original image and a cloud desktop image.
Optionally, extracting the image gradient of the denoised original image and the cloud desktop image in the step S2 includes:
based on the denoised original image and the cloud desktop image obtained in the step S2, calculating the gradient of the image by using the image edge response, wherein the image edge response calculation flow is as follows:
s21: constructing an image edge detection operator in any direction:
wherein ,
an angle representing an edge of the image; />
Is the pixel coordinate position; />
and />
The differential operators along the x axis and the y axis are calculated by the following modes:
wherein ,
variance as gaussian function; e is a natural constant;
s22: calculating an edge response of the image:
wherein ,
and />
Edge response along the x-axis and y-axis, respectively:
wherein ,
representing a convolution operation; />
，/>
and />
The images are respectively the original image after denoising and the cloud desktop image;
s23: edge response of the composite image calculates gradients of the image:
wherein ,
for the angle class of the image edge +.>
Are respectively->
、/>
、/>
And
；/>
is->
Edge response under angle; />
，/>
and />
The gradient of the image of the original image after denoising and the image of the cloud desktop end are respectively obtained.
Optionally, in the step S3, a super-resolution reconstruction network is constructed based on the denoised original image in the step S1, the cloud desktop image and the image extracted in the step S2, and the method includes:
inputting the denoised original image obtained in the step S1, the cloud desktop end image and the gradient extracted in the step S2 into a super-resolution reconstruction network, wherein the flow of the super-resolution reconstruction network is as follows:
s31: defining the output of the super-resolution reconstruction network:
wherein SR is the super-resolution reconstruction network,
and />
Respectively reconstructing the weight and the bias of the network for super resolution; />
and />
Respectively obtaining cloud desktop end images with improved definition and gradients thereof;
s32: calculating a loss function of the super-resolution reconstruction network:
the loss function of the super-resolution reconstruction network consists of two parts, wherein the first part calculates the error between the cloud desktop end image with improved definition and the original image after denoising:
wherein H, W and C are the height, width and dimension of the image respectively;
，
，/>
；
the second part calculates the error between the gradient of the cloud desktop end image with improved definition and the gradient of the original image after denoising:
the errors of the two parts are combined to form a loss function of the super-resolution reconstruction network:
wherein ,
is a balance coefficient.
Optionally, training the network constructed in step S3 in step S4, and obtaining the cloud desktop image with improved definition based on the super-resolution reconstructed network after training, including:
the parameter updating targets of the super-resolution reconstruction network are as follows:
wherein ,
and />
Respectively obtaining the weight and the bias of the super-resolution reconstruction network after updating;
representing acquisition minimization->
Reconstructing parameters of a network by using time super-resolution;
the parameter updating mode is as follows:
wherein ,
；/>
for momentum coefficients, the control parameter update depends on the specific gravity of the last update; />
For learning rate, controlling the amplitude of parameter update; t is the current update times; />
The size of the parameter update for the t time; />
Representing the partial derivative of the loss function to the super-resolution reconstruction network parameters;
after the super-resolution reconstruction network training is completed, the denoised cloud desktop end image is input into the super-resolution reconstruction network to obtain a cloud desktop end image with improved definition:
wherein ,
and />
The method comprises the steps of respectively obtaining cloud desktop end images with improved definition and corresponding image gradients.
Optionally, in the step S5, dynamic range adjustment is performed on the cloud desktop end image with improved definition, so as to obtain a cloud desktop end image with enhanced final image quality, including:
the cloud desktop end image with improved definition is subjected to dynamic range adjustment according to the dynamic range adjustment interval, and the calculation method of the dynamic range adjustment comprises the following steps:
wherein ,
is a dynamic adjustment range; />
The cloud desktop end image with improved definition is obtained; />
And (5) obtaining the cloud desktop end image with enhanced final image quality.
The invention also provides an image quality enhancement system based on the cloud desktop, which comprises:
and the image acquisition and denoising module: collecting an original image and a cloud desktop end image transmitted to a cloud desktop, and denoising the image;
an image gradient extraction module: extracting image gradients of the denoised original image and the cloud desktop end image
And a super-resolution reconstruction module: obtaining a cloud desktop end image with improved definition based on the super-resolution reconstruction network after training;
and a dynamic adjustment module: and carrying out dynamic range adjustment on the cloud desktop end image with improved definition.
Advantageous effects
Through the steps of denoising, super-resolution reconstruction, dynamic range adjustment and the like, the method can improve the definition of the cloud desktop end image. The denoising operation removes noise in the image, the super-resolution reconstruction utilizes the local features of the image to increase the details and resolution of the image, and the dynamic range is adjusted to improve the contrast and detail expressive force of the image, so that the image is clearer and more vivid.
By enhancing the image quality of the cloud desktop end image, the user can enjoy better visual experience. The definition of the image is improved, the visual effect is enhanced, the image is clearer and finer, the details are clearer and more visible, and therefore the perception and understanding of the user on the image content are improved.
Through the steps of denoising, super-resolution reconstruction, dynamic range adjustment and the like on the image at the cloud desktop end, the improvement of the image quality of the image is ensured, and meanwhile, the burden of data transmission and processing is not excessively increased, so that the method is suitable for the real-time requirement of the cloud desktop environment.
In summary, the image quality enhancement method based on the cloud desktop provides better image display effects by improving image definition, improving visual experience, improving application usability, maintaining transmission and processing efficiency, providing adaptability and customizable aspects, and the like.
Drawings
Fig. 1 is a flowchart illustrating a method for enhancing image quality based on a cloud desktop according to an embodiment of the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings, without limiting the invention in any way, and any alterations or substitutions based on the teachings of the invention are intended to fall within the scope of the invention.
Example 1: an image quality enhancement method based on a cloud desktop, as shown in fig. 1, comprises the following steps:
s1: the method comprises the steps of obtaining an original image and a cloud desktop end image transmitted to a cloud desktop, denoising the image, and obtaining the denoised original image and the cloud desktop end image.
The method comprises the steps of obtaining an original image and a cloud desktop end image transmitted to a cloud desktop, wherein the obtained image data are as follows:
wherein ,
and />
Respectively obtaining an ith original image and a cloud desktop end image;
denoising an original image and a cloud desktop end image based on self-adaptive space-time median filtering, wherein the self-adaptive space-time median filtering comprises the following steps:
s11: calculating the adaptive filter window size:
the size represents the size of the filtering window, in this embodiment, 25 th is a window screening threshold, and the calculation mode is as follows:
wherein ,
，/>
，/>
respectively the maximum value, the minimum value and the average value of the ith image;
for the i-1 th imageThe average value;
s12: calculating a space-time median filtering result:
order the
For the median filter window of the ith image,
for the median filter window of the i-1 th image, the median filter results are:
wherein ,
calculating the median value of the input sequence by the function; />
，/>
and />
The method comprises the steps of respectively obtaining an original image and a cloud desktop image.
Denoising can improve the image quality and lay a foundation for subsequent processing. Noise existing in the image can cause interference to image content and details, denoising can restore clearer and continuous edges and details of the original image, higher-quality input is provided for subsequent image processing steps such as feature extraction, image matching and the like, and more accurate processing results are facilitated to be obtained.
The two denoised images are closer in visual effect, which is beneficial to comparison and fusion between the two images. Denoising can reduce image quality degradation caused by transmission, so that the cloud desktop end image is more similar to the original image, and a more consistent basis is provided for directly comparing pixels and the like.
The denoised image can better reflect the real scene and provide more accurate image content. The image noise can distort the image, particularly noise generated in the image compression and transmission process can distort the image content and details, denoising can restore the real information of the image to a certain extent, the image content can reflect the real scene more accurately, and the method is very important for image analysis tasks such as content identification and understanding.
S2: and extracting the image gradient of the denoised original image and the cloud desktop end image.
Based on the denoised original image and the cloud desktop image obtained in the step S2, calculating the gradient of the image by using the image edge response, wherein the image edge response calculation flow is as follows:
s21: constructing an image edge detection operator in any direction:
wherein ,
an angle representing an edge of the image; />
Is the pixel coordinate position; />
and />
The differential operators along the x axis and the y axis are calculated by the following modes:
wherein ,
variance of Gaussian function, in this embodiment +.>
The method comprises the steps of carrying out a first treatment on the surface of the e is a natural constant;
s22: calculating an edge response of the image:
wherein ,
and />
Edge response along the x-axis and y-axis, respectively:
wherein ,
representing a convolution operation; />
，/>
and />
The images are respectively the original image after denoising and the cloud desktop image;
s23: edge response of the composite image calculates gradients of the image:
wherein ,
for the angle class of the image edge +.>
Are respectively->
、/>
、/>
And
；/>
is->
Edge response under angle; />
，/>
and />
The gradient of the image of the original image after denoising and the image of the cloud desktop end are respectively obtained.
Compared with the pixel value, the gradient characteristic has lower sensitivity to image blurring and compression, and can retain the image structure information lost in the image compression transmission process to a certain extent. The extracted gradient features are more stable, and the method is more suitable for judging the corresponding relation between the original image and the cloud desktop end image.
By analyzing the difference of gradient characteristics of the original image and the cloud desktop image, the areas of the image can be determined to lose more information in the transmission process, and the type of the lost information can be determined, so that guidance can be provided for designing an enhancement scheme for the cloud desktop image.
S3: and (3) constructing a super-resolution reconstruction network based on the original image after denoising in the step (S1), the cloud desktop end image and the gradient extracted in the step (S2).
Inputting the denoised original image obtained in the step S1, the cloud desktop end image and the image extracted in the step S2 into a super-resolution reconstruction network, wherein the flow of the super-resolution reconstruction network is as follows:
s31: defining the output of the super-resolution reconstruction network:
wherein SR is the super-resolution reconstruction network,
and />
Respectively reconstructing the weight and the bias of the network for super resolution; />
and />
Respectively obtaining cloud desktop end images with improved definition and gradients thereof;
s32: calculating a loss function of the super-resolution reconstruction network:
the loss function of the super-resolution reconstruction network consists of two parts, wherein the first part calculates the error between the cloud desktop end image with improved definition and the original image after denoising:
wherein H, W and C are the height, width and dimension of the image respectively;
，/>
，/>
；
the second part calculates the error between the gradient of the cloud desktop end image with improved definition and the gradient of the original image after denoising:
the errors of the two parts are combined to form a loss function of the super-resolution reconstruction network:
wherein ,
for the balance coefficient, 0.8 in this embodiment.
The resolution of the original image is higher, and the original image contains more abundant image details, so that information support is provided for obtaining a high-quality super-resolution reconstruction result. The super-resolution network can migrate the detail information to the cloud desktop end image through learning, so that resolution improvement and detail recovery are realized.
During image compression and transmission, a large loss of pixel information may occur, while gradient features may preserve structural information of the image to some extent. The network can obtain more comprehensive and accurate image representation by combining pixel information and gradient characteristics, which is beneficial to learning the restoration transformation of the image.
S4: training the network constructed in the step S3, and obtaining the cloud desktop end image with improved definition based on the super-resolution reconstruction network after training.
The parameter updating targets of the super-resolution reconstruction network are as follows:
wherein ,
and />
Respectively obtaining the weight and the bias of the super-resolution reconstruction network after updating;
representing acquisition minimization->
Reconstructing parameters of a network by using time super-resolution;
the parameter updating mode is as follows:
wherein ,
；/>
for the momentum coefficient, the control parameter update depends on the specific gravity of the last update, which is 0.8 in this embodiment; />
For learning rate, the magnitude of parameter update is controlled, in this embodiment, to be 0.001; t is the current update times; />
The size of the parameter update for the t time; />
Representing the partial derivative of the loss function to the super-resolution reconstruction network parameters;
after the super-resolution reconstruction network training is completed, the denoised cloud desktop end image is input into the super-resolution reconstruction network to obtain a cloud desktop end image with improved definition:
wherein ,
and />
The method comprises the steps of respectively obtaining cloud desktop end images with improved definition and corresponding image gradients.
S5: and carrying out dynamic range adjustment on the cloud desktop end image with improved definition to obtain the cloud desktop end image with enhanced final image quality.
The cloud desktop end image with improved definition is subjected to dynamic range adjustment according to the dynamic range adjustment interval, and the calculation method of the dynamic range adjustment comprises the following steps:
wherein ,
is a dynamic adjustment range; />
The cloud desktop end image with improved definition is obtained; />
And (5) obtaining the cloud desktop end image with enhanced final image quality.
Super-resolution reconstruction networks, while improving image sharpness and detail, also reduce the dynamic range of the image compared to the original image, which reduces the overall contrast of the image, resulting in insufficient brightness and reduced image quality. The dynamic range adjustment can be used for carrying out contrast stretching on the super-resolution image, so that the dynamic range of the super-resolution image is close to that of the original image, and the visual effect of the image is obviously improved.
Dynamic range adjustment is a simple and efficient means of enhancing the visual quality of an image as an image post-processing tool. It is easier to implement and apply than complex image operations, but can also bring about a significant image quality improvement, which makes it very suitable for use in image enhancement systems.
Example 2: the invention also discloses an image quality enhancement system based on the cloud desktop, which comprises the following four modules:
and the image acquisition and denoising module: collecting an original image and a cloud desktop end image transmitted to a cloud desktop, and denoising the image;
an image gradient extraction module: extracting image gradients of the denoised original image and the cloud desktop end image
And a super-resolution reconstruction module: obtaining a cloud desktop end image with improved definition based on the super-resolution reconstruction network after training;
and a dynamic adjustment module: and carrying out dynamic range adjustment on the cloud desktop end image with improved definition.
It should be noted that, the foregoing reference numerals of the embodiments of the present invention are merely for describing the embodiments, and do not represent the advantages and disadvantages of the embodiments. And the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, apparatus, article, or method. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, apparatus, article or method that comprises the element.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) as described above, comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the method according to the embodiments of the present invention.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (7)
1. The image quality enhancement method based on the cloud desktop is characterized by comprising the following steps of:
s1: acquiring an original image and a cloud desktop end image transmitted to a cloud desktop, and denoising the image to obtain the denoised original image and the cloud desktop end image;
s2: extracting image gradients of the denoised original image and the cloud desktop end image;
s3: constructing a super-resolution reconstruction network based on the original image after denoising in the step S1, the cloud desktop end image and the image extracted in the step S2;
s4: training the network constructed in the step S3, and obtaining a cloud desktop end image with improved definition based on the super-resolution reconstruction network after training;
s5: and carrying out dynamic range adjustment on the cloud desktop end image with improved definition to obtain the cloud desktop end image with enhanced final image quality.
2. The cloud desktop-based image quality enhancement method according to claim 1, wherein the step S1 includes:
the method comprises the steps of obtaining an original image and a cloud desktop end image transmitted to a cloud desktop, wherein the obtained image data are as follows:
wherein ,
and />
Respectively the acquired firstiAn original image and a cloud desktop end image;
denoising an original image and a cloud desktop end image based on self-adaptive space-time median filtering, wherein the self-adaptive space-time median filtering comprises the following steps:
s11: calculating the adaptive filter window size:
wherein ,sizethe size of the filter window is indicated,ththe window screening threshold is calculated by the following steps:
wherein ,
，/>
，/>
respectively the firstiMaximum, minimum and mean values of the images;
is the firsti-a mean of 1 image;
s12: calculating a space-time median filtering result:
order the
Is the firstiThe median filter window of the sheet of image,
is the firsti-median filter window of 1 image, then the result of median filtering is: />
；
wherein ,
calculating the median value of the input sequence by the function; />
，/>
and />
The method comprises the steps of respectively obtaining an original image and a cloud desktop image.
3. The cloud desktop-based image quality enhancement method according to claim 2, wherein the step S2 includes:
based on the denoised original image and the cloud desktop image obtained in the step S2, calculating the gradient of the image by using the image edge response, wherein the image edge response calculation flow is as follows:
s21: constructing an image edge detection operator in any direction:
wherein ,
an angle representing an edge of the image; />
Is the pixel coordinate position; />
and />
Respectively the edgesxShaft and method for producing the sameyThe differential operator of the shaft is calculated by the following steps:
wherein ,
variance as gaussian function;eis a natural constant;
s22: calculating an edge response of the image:
wherein ,
and />
Respectively the edgesxShaft and method for producing the sameyEdge response of the shaft:
wherein ,
representing a convolution operation; />
，/>
and />
The images are respectively the original image after denoising and the cloud desktop image;
s23: edge response of the composite image calculates gradients of the image:
wherein ,
for the angle class of the image edge +.>
Are respectively->
、/>
、/>
and />
；
Is->
Edge response under angle; />
，/>
and />
The gradient of the image of the original image after denoising and the image of the cloud desktop end are respectively obtained.
4. The cloud desktop-based image quality enhancement method according to claim 3, wherein in the step S3, it includes:
inputting the denoised original image obtained in the step S1, the cloud desktop end image and the gradient extracted in the step S2 into a super-resolution reconstruction network, wherein the flow of the super-resolution reconstruction network is as follows:
s31: defining the output of the super-resolution reconstruction network:
wherein ,SRfor the super-resolution reconstruction of the network,
and />
Respectively reconstructing the weight and the bias of the network for super resolution;
and />
Respectively obtaining cloud desktop end images with improved definition and gradients thereof;
s32: calculating a loss function of the super-resolution reconstruction network:
the loss function of the super-resolution reconstruction network is composed of two parts, wherein the first part calculates the error between the cloud desktop end image with improved definition and the original image after denoising:
wherein ,H，WandCthe height, width and dimension of the image respectively;
，/>
，
；
the second part calculates the error between the gradient of the cloud desktop end image with improved definition and the gradient of the original image after denoising:
the errors of the two parts are combined to form a loss function of the super-resolution reconstruction network:
wherein ,
is a balance coefficient.
5. The cloud desktop-based image quality enhancement method according to claim 4, wherein in the step S4, it includes:
the parameter updating targets of the super-resolution reconstruction network are as follows:
wherein ,
and />
Respectively obtaining the weight and the bias of the super-resolution reconstruction network after updating; />
Representing acquisition minimization->
Reconstructing parameters of a network by using time super-resolution;
the parameter updating mode is as follows:
wherein ,
；/>
for momentum coefficients, the control parameter update depends on the specific gravity of the last update;
for learning rate, controlling the amplitude of parameter update;tthe current update times; />
Is the firsttThe size of the secondary parameter update;
representing the partial derivative of the loss function to the super-resolution reconstruction network parameters;
after the super-resolution reconstruction network training is completed, the denoised cloud desktop end image is input into the super-resolution reconstruction network to obtain a cloud desktop end image with improved definition:
wherein ,
and />
The method comprises the steps of respectively obtaining cloud desktop end images with improved definition and corresponding image gradients.
6. The cloud desktop-based image quality enhancement method according to claim 5, wherein in the step S5, it includes:
the cloud desktop end image with improved definition is subjected to dynamic range adjustment according to the dynamic range adjustment interval, and the calculation method of the dynamic range adjustment comprises the following steps:
wherein ,
is a dynamic adjustment range; />
The cloud desktop end image with improved definition is obtained; />
And (5) obtaining the cloud desktop end image with enhanced final image quality.
7. An image quality enhancement system based on a cloud desktop, comprising:
and the image acquisition and denoising module: collecting an original image and a cloud desktop end image transmitted to a cloud desktop, and denoising the image;
an image gradient extraction module: extracting image gradients of the denoised original image and the cloud desktop end image
And a super-resolution reconstruction module: obtaining a cloud desktop end image with improved definition based on the super-resolution reconstruction network after training;
and a dynamic adjustment module: carrying out dynamic range adjustment on the cloud desktop end image with improved definition;
to realize the cloud desktop-based image quality enhancement method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310661815.3A CN116385318B (en) | 2023-06-06 | 2023-06-06 | Image quality enhancement method and system based on cloud desktop |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310661815.3A CN116385318B (en) | 2023-06-06 | 2023-06-06 | Image quality enhancement method and system based on cloud desktop |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116385318A true CN116385318A (en) | 2023-07-04 |
| CN116385318B CN116385318B (en) | 2023-10-10 |
Family
ID=86969794
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310661815.3A Active CN116385318B (en) | 2023-06-06 | 2023-06-06 | Image quality enhancement method and system based on cloud desktop |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116385318B (en) |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4973111A (en) * | 1988-09-14 | 1990-11-27 | Case Western Reserve University | Parametric image reconstruction using a high-resolution, high signal-to-noise technique |
| KR20140040322A (en) * | 2012-09-24 | 2014-04-03 | 재단법인대구경북과학기술원 | Single image super-resolution image reconstruction device and method thereof |
| US20200249314A1 (en) * | 2019-02-01 | 2020-08-06 | GM Global Technology Operations LLC | Deep learning for super resolution in a radar system |
| CN111696033A (en) * | 2020-05-07 | 2020-09-22 | 中山大学 | Real image super-resolution model and method for learning cascaded hourglass network structure based on angular point guide |
| CN112733894A (en) * | 2020-12-29 | 2021-04-30 | 广东省电信规划设计院有限公司 | Intelligent video identification sensing method based on wireless communication technology |
| WO2021107290A1 (en) * | 2019-11-28 | 2021-06-03 | Samsung Electronics Co., Ltd. | Electronic apparatus and controlling method thereof |
| WO2022057837A1 (en) * | 2020-09-16 | 2022-03-24 | 广州虎牙科技有限公司 | Image processing method and apparatus, portrait super-resolution reconstruction method and apparatus, and portrait super-resolution reconstruction model training method and apparatus, electronic device, and storage medium |
| WO2022116933A1 (en) * | 2020-12-04 | 2022-06-09 | 华为技术有限公司 | Model training method, data processing method and apparatus |
| CN114841308A (en) * | 2022-03-17 | 2022-08-02 | 阿里巴巴（中国）有限公司 | Super-resolution reconstruction method, device and equipment for cloud desktop image and storage medium |
| US20230052483A1 (en) * | 2020-02-17 | 2023-02-16 | Intel Corporation | Super resolution using convolutional neural network |
| WO2023035531A1 (en) * | 2021-09-10 | 2023-03-16 | 平安科技（深圳）有限公司 | Super-resolution reconstruction method for text image and related device thereof |
| CN115941965A (en) * | 2022-12-09 | 2023-04-07 | 阿里巴巴（中国）有限公司 | Cloud desktop coding method, reconstruction method, display method and display system |
-
2023
- 2023-06-06 CN CN202310661815.3A patent/CN116385318B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4973111A (en) * | 1988-09-14 | 1990-11-27 | Case Western Reserve University | Parametric image reconstruction using a high-resolution, high signal-to-noise technique |
| KR20140040322A (en) * | 2012-09-24 | 2014-04-03 | 재단법인대구경북과학기술원 | Single image super-resolution image reconstruction device and method thereof |
| US20200249314A1 (en) * | 2019-02-01 | 2020-08-06 | GM Global Technology Operations LLC | Deep learning for super resolution in a radar system |
| WO2021107290A1 (en) * | 2019-11-28 | 2021-06-03 | Samsung Electronics Co., Ltd. | Electronic apparatus and controlling method thereof |
| US20230052483A1 (en) * | 2020-02-17 | 2023-02-16 | Intel Corporation | Super resolution using convolutional neural network |
| CN111696033A (en) * | 2020-05-07 | 2020-09-22 | 中山大学 | Real image super-resolution model and method for learning cascaded hourglass network structure based on angular point guide |
| WO2022057837A1 (en) * | 2020-09-16 | 2022-03-24 | 广州虎牙科技有限公司 | Image processing method and apparatus, portrait super-resolution reconstruction method and apparatus, and portrait super-resolution reconstruction model training method and apparatus, electronic device, and storage medium |
| WO2022116933A1 (en) * | 2020-12-04 | 2022-06-09 | 华为技术有限公司 | Model training method, data processing method and apparatus |
| CN112733894A (en) * | 2020-12-29 | 2021-04-30 | 广东省电信规划设计院有限公司 | Intelligent video identification sensing method based on wireless communication technology |
| WO2023035531A1 (en) * | 2021-09-10 | 2023-03-16 | 平安科技（深圳）有限公司 | Super-resolution reconstruction method for text image and related device thereof |
| CN114841308A (en) * | 2022-03-17 | 2022-08-02 | 阿里巴巴（中国）有限公司 | Super-resolution reconstruction method, device and equipment for cloud desktop image and storage medium |
| CN115941965A (en) * | 2022-12-09 | 2023-04-07 | 阿里巴巴（中国）有限公司 | Cloud desktop coding method, reconstruction method, display method and display system |
Non-Patent Citations (2)
| Title |
|---|
| G. SEIF: ""Edge-Based Loss Function for Single Image Super-Resolution"", 《ICASSP》, pages 1 - 5 * |
| 孟志青: ""监督损失函数光滑化图像超分辨率重建"", 《中国图象图形学报》, pages 1 - 12 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116385318B (en) | 2023-10-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111539879B (en) | Video blind denoising method and device based on deep learning | |
| CN107123089B (en) | Remote sensing image super-resolution reconstruction method and system based on depth convolution network | |
| Dong et al. | Nonlocal back-projection for adaptive image enlargement | |
| CN110163818B (en) | Low-illumination video image enhancement method for maritime unmanned aerial vehicle | |
| JPH11150669A (en) | Image improving device | |
| CN111667410B (en) | Image resolution improving method and device and electronic equipment | |
| CN110827397B (en) | Texture fusion method for real-time three-dimensional reconstruction of RGB-D camera | |
| CN110796616B (en) | Turbulence degradation image recovery method based on norm constraint and self-adaptive weighted gradient | |
| WO2014070273A1 (en) | Recursive conditional means image denoising | |
| CN109003233B (en) | Image denoising method based on self-adaptive weight total variation model | |
| CN116385318B (en) | Image quality enhancement method and system based on cloud desktop | |
| Raveendran et al. | Image fusion using LEP filtering and bilinear interpolation | |
| CN112819739A (en) | Scanning electron microscope image processing method and system | |
| CN112509144A (en) | Face image processing method and device, electronic equipment and storage medium | |
| CN117333359A (en) | Mountain-water painting image super-resolution reconstruction method based on separable convolution network | |
| CN115965552B (en) | Frequency-space-time domain joint denoising and recovering system for low signal-to-noise ratio image sequence | |
| CN115116468A (en) | Video generation method and device, storage medium and electronic equipment | |
| CN115082296A (en) | Image generation method based on wavelet domain image generation framework | |
| Son et al. | A pair of noisy/blurry patches-based PSF estimation and channel-dependent deblurring | |
| Tun et al. | Joint Training of Noisy Image Patch and Impulse Response of Low-Pass Filter in CNN for Image Denoising | |
| Zheng et al. | Regularization parameter selection for total variation model based on local spectral response | |
| CN111724332B (en) | Image enhancement method and system suitable for closed cavity detection | |
| CN112435192B (en) | Lightweight image definition enhancing method | |
| CN108510449B (en) | Total variation image noise elimination method based on self-adaptive kernel regression | |
| CN116721015A (en) | Noise-containing rapid image enhancement and super-resolution reconstruction method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |