CN117611548A - Image quality evaluation method and system based on distortion information - Google Patents
Image quality evaluation method and system based on distortion information Download PDFInfo
- Publication number
- CN117611548A CN117611548A CN202311577129.4A CN202311577129A CN117611548A CN 117611548 A CN117611548 A CN 117611548A CN 202311577129 A CN202311577129 A CN 202311577129A CN 117611548 A CN117611548 A CN 117611548A
- Authority
- CN
- China
- Prior art keywords
- image
- distortion
- extraction module
- input
- quality evaluation
- 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.)
- Pending
Links
- 238000013441 quality evaluation Methods 0.000 title claims abstract description 52
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000000605 extraction Methods 0.000 claims abstract description 74
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 32
- 238000013210 evaluation model Methods 0.000 claims abstract description 21
- 238000010606 normalization Methods 0.000 claims abstract description 15
- 238000012545 processing Methods 0.000 claims abstract description 5
- 238000007781 pre-processing Methods 0.000 claims description 17
- 238000011176 pooling Methods 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 5
- 238000001303 quality assessment method Methods 0.000 claims description 4
- 230000009191 jumping Effects 0.000 claims description 3
- 239000013598 vector Substances 0.000 claims description 3
- 230000006870 function Effects 0.000 description 7
- 230000006835 compression Effects 0.000 description 4
- 238000007906 compression Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000013135 deep learning Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 230000008447 perception Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000004069 differentiation Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 1
- 238000009792 diffusion process Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013209 evaluation strategy Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 238000005562 fading Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a distortion-based image quality evaluation method and a distortion-based image quality evaluation system, wherein an input image is preprocessed to obtain a plurality of input image blocks; then inputting the image block into a distortion-based image quality evaluation model to obtain an overall quality score; the image evaluation model based on the distortion information comprises a distortion extraction module, a feature extraction module and a weighted summation module; the feature extraction module adopts a convolutional neural network S and a normalization layer, wherein the input of the feature extraction module is a preprocessed image, and the output of the feature extraction module is a local attention weight; the distortion extraction module comprises a distortion extraction network and a convolutional neural network W, wherein the input of the distortion extraction module is a preprocessed image, and the output of the distortion extraction module is a local mass fraction; the weighted summation module obtains the overall quality score of the image by weighted summation of the local quality score and the local attention weight. The invention has guiding significance for guiding image processing tasks such as image restoration, denoising, reconstruction and the like.
Description
Technical Field
The present invention relates to the field of image quality processing technologies, and in particular, to an image quality evaluation method and system based on distortion information.
Background
The advent of the media era has brought about explosive growth of information such as images and videos. The difference in image quality is large due to the difference in the acquisition manner of these pieces of information at the time of acquisition (document 1). In such a background, it becomes particularly important how to accurately evaluate the image quality. By means of better image quality evaluation methods and strategies, better quality images can be screened out, and attractive visual experience is presented for information in the fused media era.
Image quality evaluation can be classified into Full Reference (FR), partial reference (RR), and No Reference (NR) according to availability of reference image information (document 2). FR-IQA needs to be evaluated by comparing the difference between the image to be measured and the reference image, and RR-IQA needs to be evaluated by partial information of the reference image. In the practical application process, reference information cannot be obtained, so that NR-IQA becomes particularly wide-ranging in practical application.
With the development of machine learning, shallow machine learning and deep machine learning are introduced successively, and the general image quality assessment method has a long-standing development. Most of these methods use CNN feature extraction to predict the type of image distortion and regress to obtain a quality score according to the distortion type.
Considering the complexity of the distortion type, the subsequent image quality evaluation method based on deep learning mostly removes the prediction of the distortion type, and converts the method into the method of directly obtaining the image quality score through feature extraction and linear regression. For example, waDIQaM (document 3) obtains a no-reference quality evaluation by training a full-reference image quality evaluation model and sharing parameters to a no-reference image quality evaluation model and further performing transfer learning. In addition, the WaDIQaM method introduces an attention mechanism, and the perception difference of human eyes on different parts of the image is simulated by distributing different attention weights to the different parts of the image, so that the quality evaluation provided by the model can better accord with the subjective feeling of the human eyes.
While CNNs can extract various features of images, current research generally suggests that: there is a large amount of information redundancy in feature extraction of CNN (document 4). Therefore, how to better realize screening and multiplexing of features and enhance the interpretability of CNN feature extraction is an important research topic of current image quality evaluation research.
[1]Wang Z.Applications of objective image quality assessment methods[applications corner][J].IEEE signal processing magazine,2011,28(6):137-142.
[2]Wang Z,Bovik AC,Sheikh H R,et al.Image quality assessment:from error visibility to structural similarity[J].IEEE transactions on image processing,2004,13(4):600-612.
[3]Bosse S,Maniry D,Müller K R,et al.Deep neural networks for no-reference and full-reference image quality assessment[J].IEEE Transactions on image processing,2017,27(1):206-219.
[4]Zhang Q,Jiang Z,Lu Q,et al.Split to be slim:An overlooked redundancy in vanilla convolution[J].arXiv preprint arXiv:2006.12085,2020.
Disclosure of Invention
In order to screen and multiplex the image features extracted by CNN, the invention provides an image quality evaluation method and system based on distortion information. The method realizes that the image quality is reflected by only extracting the characteristics of the image distortion information through the image quality evaluation model based on the distortion information, and provides a new direction for image quality evaluation research. Meanwhile, a low-cost training mode is also provided, and the utilization efficiency of the features is greatly improved.
The main technical scheme of the invention is as follows: an image quality evaluation method based on distortion information includes the steps of:
step 1, preprocessing an input image and dividing the input image into a plurality of image blocks;
step 2, inputting the preprocessed image blocks into an image evaluation model based on distortion information to obtain the overall quality fraction of the image;
the image evaluation model based on the distortion information comprises a distortion extraction module, a feature extraction module and a weighted summation module;
the feature extraction module adopts a convolutional neural network S and a normalization layer, wherein the input of the feature extraction module is a preprocessed image, and the output of the feature extraction module is a local attention weight;
the distortion extraction module comprises a distortion extraction network and a convolutional neural network W, wherein the input of the distortion extraction module is a preprocessed image, and the output of the distortion extraction module is a local mass fraction;
the weighted summation module obtains the overall quality score of the image by weighted summation of the local quality score and the local attention weight.
Further, in step 1, preprocessing is performed on the input image, including randomly cutting the input image into a plurality of image blocks, and normalizing the image blocks, where a specific normalization function is as follows:
where x is the input image block, x n Is the normalized input image block.
Further, in step 2, the convolutional neural network S includes 5 convolutional blocks and one linear layer, where the first two convolutional layers include two 3×3 convolutional kernels and one max pooling layer; the last three convolution layers comprise two 3 x 3 convolution kernels, one 1 x 1 convolution kernel and one maximum pooling layer to output distortion information with the same input width, height and dimension; the number of input channels of the linear layer is the characteristic dimension, and the number of output channels is the number of image blocks.
Further, the distortion extraction network consists of three full convolution networks of downsampling, convolution and upsampling, wherein the downsampling part comprises 3 convolution layers, each part comprises 3 convolution kernels of 3 multiplied by 3 and 3 maximum pooling layers, and the downsampling and the upsampling are connected in a jumping manner; the convolutional neural network W is the same as the convolutional neural network S architecture in the feature extraction module.
Further, the image evaluation model in the step 2 is a trained multi-scale convolutional neural network model; the training process comprises the following substeps:
step S1, importing an image quality evaluation data set;
s2, preprocessing the data;
and step S3, inputting the preprocessed data into an image evaluation model, and optimizing parameters of the model by a method of gradient descent and minimizing a loss function, so that the image evaluation model outputs the quality fraction of the input image more accurately.
Further, the data set in step S1 includes a synthetic distortion data set and a true distortion data set.
Further, the preprocessing in step S2 includes random clipping and normalization processing.
Further, the specific expression of the loss function in step S3 is as follows:
wherein the method comprises the steps ofAnd->Is two input vectors, which respectively refer to a predicted value given by an image evaluation model and a true quality score of an image, N is the total amount of input samples, and i is an index of a plurality of sample values contained in each sample.
In a second aspect, the present invention also provides an image quality evaluation system based on distortion information, including the following units:
the preprocessing unit is used for preprocessing an input image and dividing the input image into a plurality of image blocks;
the quality evaluation unit is used for inputting the preprocessed image blocks into an image evaluation model based on distortion information to obtain the overall quality fraction of the image;
the image evaluation model based on the distortion information comprises a distortion extraction module, a feature extraction module and a weighted summation module;
the feature extraction module adopts a convolutional neural network S and a normalization layer, wherein the input of the feature extraction module is a preprocessed image, and the output of the feature extraction module is a local attention weight;
the distortion extraction module comprises a distortion extraction network and a convolutional neural network W, wherein the input of the distortion extraction module is a preprocessed image, and the output of the distortion extraction module is a local mass fraction;
the weighted summation module obtains the overall quality score of the image by weighted summation of the local quality score and the local attention weight.
In a third aspect, the present invention also provides an image quality evaluation apparatus based on distortion information, comprising:
one or more processors;
and a storage means for storing one or more programs which, when executed by the one or more processors, cause the one or more processors to implement an image quality evaluation method based on distortion information as described in the above schemes.
The invention adopts the image quality evaluation model based on the distortion image quality evaluation, and respectively obtains the local quality fraction and the local attention weight through the distortion extraction module and the characteristic extraction module, and finally, the local quality fraction and the local attention weight are weighted and summed to obtain the integral quality fraction of the input image. Compared with the traditional image quality evaluation method based on deep learning, the image quality evaluation method based on deep learning only evaluates the image quality from the distortion information, reduces information redundancy, reduces training difficulty, enables a model to reflect the distortion degree of the image more objectively and accurately, and keeps consistent with subjective perception of human eyes.
Drawings
The following examples, as well as specific embodiments, are used to further illustrate the technical solutions herein. In addition, in the course of describing the technical solutions, some drawings are also used. Other figures and the intent of the present invention can be derived from these figures without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method according to an embodiment of the present invention;
FIG. 2 is a diagram of a multi-level feature multiplexing image quality evaluation model according to an embodiment of the present invention;
fig. 3 is a training flow chart of a multi-level feature multiplexing image quality evaluation model according to an embodiment of the invention.
Detailed Description
In order to facilitate the understanding and practice of the invention, those of ordinary skill in the art will now make further details with reference to the drawings and examples, it being understood that the examples described herein are for the purpose of illustration and explanation only and are not intended to limit the invention thereto.
The present embodiment takes a given image data set to be measured as an example, and further describes the present invention. Referring to fig. 1, the image quality evaluation method based on distortion provided in this embodiment includes the following steps:
step 1: the input image is preprocessed and divided into a number of image blocks.
In one embodiment, preprocessing is performed on an input image, including randomly cropping into 32 image blocks of 32×32, and normalizing the image, where a specific normalization function is:
where x is the input image block, x n Is the normalized input image block.
This normalization strategy greatly reduces the data distribution differentiation, thereby improving the generalization performance of the model.
Step 2: inputting the preprocessed image blocks into an image evaluation model based on distortion information to obtain local quality scores and local attention weights; after the local quality score and the local attention weight are respectively obtained, the whole quality score of the image is obtained through weighted summation;
please refer to fig. 2, the image quality evaluation model based on distortion information includes a distortion extraction module, a feature extraction module and a weighted summation module;
in one embodiment, the feature extraction module adopts a convolutional neural network S and a normalization layer, the convolutional neural network S performs feature extraction on an original image to generate attention weights, and the normalization layer greatly reduces data distribution differentiation, so that generalization performance of the model is improved. Wherein the convolutional neural network S comprises 5 convolutional blocks, and one linear layer, wherein the first two convolutional layers comprise two 3 x 3 convolutional kernels and one max pooling layer; the last three convolution layers comprise two 3 x 3 convolution kernels, one 1 x 1 convolution kernel and one maximum pooling layer to output distortion information with the same input width, height and dimension; the number of input channels of the linear layer is the characteristic dimension, and the number of output channels is the number of image blocks; the input is the preprocessed image and the output is the local attention weight.
The distortion extraction module comprises a distortion extraction network and a convolutional neural network W, wherein the distortion extraction network is used for extracting distortion information of an image, and the convolutional neural network W performs feature extraction on the distortion information so as to obtain local quality fraction. The distortion extraction network consists of three full convolution networks of downsampling, convolution and upsampling, wherein the downsampling part comprises 3 convolution layers, each part comprises 3 convolution kernels with the size of 3 multiplied by 3 and 3 largest pooling layers, the downsampling and upsampling are connected in a jumping manner, and the convolution neural network W has the same architecture as the convolution neural network S in the feature extraction module; the input is the preprocessed image, and the output is the local quality fraction.
The weighted summation module obtains the overall quality score of the image by weighted summation of the local quality score and the local attention weight;
directing attention to fig. 3, in one embodiment, the image quality focus evaluation model under diffusion guidance is a trained multi-scale convolutional neural network model; the training process comprises the following substeps:
step S1: importing an image quality evaluation data set;
in one embodiment the dataset contains both a synthetic distortion dataset (e.g., LIVE, CSIQ, TID 2013) and a true distortion dataset (e.g., CLIVE, koniQ), which contain subjective evaluation scores for each image.
Step S2: preprocessing data;
in one embodiment, preprocessing is performed on an input image, including randomly cropping into 32 image blocks of 32×32, and normalizing the image, where the normalization formula is consistent with 2, so that data in different data sets are unified, and model performance is improved.
Step S3: the preprocessed data is input into a multi-scale convolutional neural network model, and parameters of the model are optimized through gradient descent and a method for minimizing a loss function, so that the model can more accurately output the quality score of an input image.
In the parameter updating process of one embodiment, the efficiency of parameter updating is improved by using the reduction of a loss function, and the loss function adopts an absolute error, which can be described as:
wherein the method comprises the steps ofAnd->Is two input vectors, respectively referring to the predicted value given by the model and the true quality score of the image, N is the total input sample amount, i is the index of the multiple sample values contained in each sample.
To verify the performance of the algorithm, we first performed experiments on LIVE, CSIQ, TID2013, clear, konIQ datasets, respectively, through four image databases to verify the performance of the algorithm. These datasets cover most of the synthetic and natural distortions. LIVE databases contain 29 reference images and 779 Zhang Shizhen images, distortion types including JPEG2000, JPEG compression, white noise, gaussian blur, and rayleigh fading, with differential mean opinion scores (Differential Mean Opinion Scores, DMOS) for each image provided in the database, with smaller DMOS values representing higher image quality. The CSIQ database contains 30 reference pictures, each with 6 distortion types, each distortion having 4-5 degrees. The TID2013 database is a widely used dataset for image quality assessment, containing 25 reference images and 3,000 Zhang Shizhen images. These distorted images cover a variety of types including gaussian noise, blurring, compression, etc. Each image is subjectively scored and the scoring value is used to quantify the perceived distortion level of the image. The clear database is a comprehensive database for image quality evaluation, containing 1,100 images, including both reference images and corresponding distorted images. Distortion types include compression, blurring, etc., each image being assigned a subjective quality score. The KonIQ database focuses on multidimensional distortion in image quality evaluation, including 1,200 images, covering various distortion types, such as blurring, compression, noise, and the like. Each image is provided with a corresponding subjective quality score, so that researchers can deeply study the influence of different distortions on the perceived quality of the image.
Table 1 shows the performance of the image quality evaluation algorithm based on distortion information and other related algorithms on LIVE, CSIQ, TID2013, clear, konIQ data sets, and it can be seen that the IQA index performs very well on all 4 databases, which represents the superior performance of the image quality evaluation algorithm based on distortion information.
TABLE 1 comparison of model Performance in different data sets
Wherein the highest and second highest SROCC and PLCC are indicated by bold and underline, respectively.
The embodiment of the invention also provides an image quality evaluation system based on the distortion information, which comprises the following units:
the preprocessing unit is used for preprocessing an input image and dividing the input image into a plurality of image blocks;
the quality evaluation unit is used for inputting the preprocessed image blocks into an image evaluation model based on distortion information to obtain the overall quality fraction of the image;
the image evaluation model based on the distortion information comprises a distortion extraction module, a feature extraction module and a weighted summation module;
the feature extraction module adopts a convolutional neural network S and a normalization layer, wherein the input of the feature extraction module is a preprocessed image, and the output of the feature extraction module is a local attention weight;
the distortion extraction module comprises a distortion extraction network and a convolutional neural network W, wherein the input of the distortion extraction module is a preprocessed image, and the output of the distortion extraction module is a local mass fraction;
the weighted summation module obtains the overall quality score of the image by weighted summation of the local quality score and the local attention weight.
The specific implementation manner of each unit is the same as that of each step, and the invention is not written.
The embodiment of the invention also provides an image quality evaluation device based on the distortion information, which comprises:
one or more processors;
storage means for storing one or more programs which, when executed by the one or more processors, cause the one or more processors to implement an image quality assessment method based on distortion information as claimed in any one of claims 1 to 8.
The invention can realize accurate quality evaluation of the image to be detected, and the evaluation result can reflect the objective distortion degree of the image and accord with subjective perception of human eyes. The method can also be applied to image processing tasks such as denoising, repairing and super-resolution restoration of the guide image, and has good popularization and application prospects.
It should be understood that the foregoing description of the preferred embodiments is not intended to limit the scope of the invention, but rather to limit the scope of the claims, and that those skilled in the art can make substitutions or modifications without departing from the scope of the invention as set forth in the appended claims.
Claims (10)
1. An image quality evaluation method based on distortion information, comprising the steps of:
step 1, preprocessing an input image and dividing the input image into a plurality of image blocks;
step 2, inputting the preprocessed image blocks into an image evaluation model based on distortion information to obtain the overall quality fraction of the image;
the image evaluation model based on the distortion information comprises a distortion extraction module, a feature extraction module and a weighted summation module;
the feature extraction module adopts a convolutional neural network S and a normalization layer, wherein the input of the feature extraction module is a preprocessed image, and the output of the feature extraction module is a local attention weight;
the distortion extraction module comprises a distortion extraction network and a convolutional neural network W, wherein the input of the distortion extraction module is a preprocessed image, and the output of the distortion extraction module is a local mass fraction;
the weighted summation module obtains the overall quality score of the image by weighted summation of the local quality score and the local attention weight.
2. The image quality evaluation method based on distortion information according to claim 1, wherein: in step 1, preprocessing an input image, including randomly cutting the input image into a plurality of image blocks, and normalizing the image blocks, wherein the specific normalization function is as follows:
wherein the method comprises the steps ofx is the input image block, x n Is the normalized input image block.
3. The image quality evaluation method based on distortion information according to claim 1, wherein: in step 2, the convolutional neural network S comprises 5 convolutional blocks and one linear layer, wherein the first two convolutional layers comprise two 3×3 convolutional kernels and one max pooling layer; the last three convolution layers comprise two 3 x 3 convolution kernels, one 1 x 1 convolution kernel and one maximum pooling layer to output distortion information with the same input width, height and dimension; the number of input channels of the linear layer is the characteristic dimension, and the number of output channels is the number of image blocks.
4. The image quality evaluation method based on distortion information according to claim 1, wherein: the distortion extraction network consists of three full convolution networks of downsampling, convolution and upsampling, wherein the downsampling part comprises 3 convolution layers, each part comprises 3 convolution kernels with the size of 3 multiplied by 3 and 3 maximum pooling layers, and the downsampling and the upsampling are connected in a jumping manner; the convolutional neural network W is the same as the convolutional neural network S architecture in the feature extraction module.
5. The image quality evaluation method based on distortion information according to claim 1, wherein: the image evaluation model in the step 2 is a trained multi-scale convolutional neural network model; the training process comprises the following substeps:
step S1, importing an image quality evaluation data set;
s2, preprocessing the data;
and step S3, inputting the preprocessed data into an image evaluation model, and optimizing parameters of the model by a method of gradient descent and minimizing a loss function, so that the image evaluation model outputs the quality fraction of the input image more accurately.
6. The image quality evaluation method based on distortion information according to claim 5, wherein: the data set in step S1 contains a synthetic distortion data set and a true distortion data set.
7. The image quality evaluation method based on distortion information according to claim 5, wherein: the preprocessing in step S2 includes random clipping and normalization processing.
8. The image quality evaluation method based on distortion information according to claim 5, wherein: the specific expression of the loss function in step S3 is as follows:
wherein the method comprises the steps ofAnd->Is two input vectors, which respectively refer to a predicted value given by an image evaluation model and a true quality score of an image, N is the total amount of input samples, and i is an index of a plurality of sample values contained in each sample.
9. An image quality evaluation system based on distortion information, comprising the following units:
the preprocessing unit is used for preprocessing an input image and dividing the input image into a plurality of image blocks;
the quality evaluation unit is used for inputting the preprocessed image blocks into an image evaluation model based on distortion information to obtain the overall quality fraction of the image;
the image evaluation model based on the distortion information comprises a distortion extraction module, a feature extraction module and a weighted summation module;
the feature extraction module adopts a convolutional neural network S and a normalization layer, wherein the input of the feature extraction module is a preprocessed image, and the output of the feature extraction module is a local attention weight;
the distortion extraction module comprises a distortion extraction network and a convolutional neural network W, wherein the input of the distortion extraction module is a preprocessed image, and the output of the distortion extraction module is a local mass fraction;
the weighted summation module obtains the overall quality score of the image by weighted summation of the local quality score and the local attention weight.
10. An image quality evaluation apparatus based on distortion information, comprising:
one or more processors;
storage means for storing one or more programs which, when executed by the one or more processors, cause the one or more processors to implement an image quality assessment method based on distortion information as claimed in any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311577129.4A CN117611548A (en) | 2023-11-22 | 2023-11-22 | Image quality evaluation method and system based on distortion information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311577129.4A CN117611548A (en) | 2023-11-22 | 2023-11-22 | Image quality evaluation method and system based on distortion information |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117611548A true CN117611548A (en) | 2024-02-27 |
Family
ID=89950873
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311577129.4A Pending CN117611548A (en) | 2023-11-22 | 2023-11-22 | Image quality evaluation method and system based on distortion information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117611548A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110060236A (en) * | 2019-03-27 | 2019-07-26 | 天津大学 | Stereo image quality evaluation method based on depth convolutional neural networks |
| CN112508967A (en) * | 2020-12-04 | 2021-03-16 | 武汉大学 | Image quality evaluation method and system |
| CN113313682A (en) * | 2021-05-28 | 2021-08-27 | 西安电子科技大学 | No-reference video quality evaluation method based on space-time multi-scale analysis |
| CN114758018A (en) * | 2020-12-25 | 2022-07-15 | 中国科学院沈阳自动化研究所 | Infrared spectrum reconstruction method based on convolutional neural network |
-
2023
- 2023-11-22 CN CN202311577129.4A patent/CN117611548A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110060236A (en) * | 2019-03-27 | 2019-07-26 | 天津大学 | Stereo image quality evaluation method based on depth convolutional neural networks |
| CN112508967A (en) * | 2020-12-04 | 2021-03-16 | 武汉大学 | Image quality evaluation method and system |
| CN114758018A (en) * | 2020-12-25 | 2022-07-15 | 中国科学院沈阳自动化研究所 | Infrared spectrum reconstruction method based on convolutional neural network |
| CN113313682A (en) * | 2021-05-28 | 2021-08-27 | 西安电子科技大学 | No-reference video quality evaluation method based on space-time multi-scale analysis |
Non-Patent Citations (1)
| Title |
|---|
| JINWEI LIU等: "A No-Reference Image Quality Assessment Methodology Based on Distortion Information Extraction", 2023 INTERNATIONAL CONFERENCE ON COMPUTER SCIENCE AND AUTOMATION TECHNOLOGY (CSAT), 8 October 2023 (2023-10-08), pages 1 - 3 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110059586B (en) | Iris positioning and segmenting system based on cavity residual error attention structure | |
| CN112633382B (en) | Method and system for classifying few sample images based on mutual neighbor | |
| CN106097360A (en) | A kind of strip steel surface defect identification method and device | |
| CN110223304B (en) | Image segmentation method and device based on multipath aggregation and computer-readable storage medium | |
| CN110263215B (en) | Video emotion positioning method and system | |
| CN114612714B (en) | Curriculum learning-based reference-free image quality evaluation method | |
| CN111488917A (en) | Garbage image fine-grained classification method based on incremental learning | |
| CN109376787B (en) | Manifold learning network and computer vision image set classification method based on manifold learning network | |
| CN113066065B (en) | No-reference image quality detection method, system, terminal and medium | |
| CN110689523A (en) | Personalized image information evaluation method based on meta-learning and information data processing terminal | |
| CN104077742B (en) | Human face sketch synthetic method and system based on Gabor characteristic | |
| CN110738660A (en) | Spine CT image segmentation method and device based on improved U-net | |
| CN117237599A (en) | Image target detection method and device | |
| CN107133579A (en) | Based on CSGF (2D)2The face identification method of PCANet convolutional networks | |
| CN112560668B (en) | Human behavior recognition method based on scene priori knowledge | |
| CN116993639A (en) | Visible light and infrared image fusion method based on structural re-parameterization | |
| CN111652238B (en) | Multi-model integration method and system | |
| CN108280417A (en) | A kind of finger vena method for quickly identifying | |
| CN112633301A (en) | Traditional Chinese medicine tongue image greasy feature classification method based on depth metric learning | |
| CN114841887B (en) | Image recovery quality evaluation method based on multi-level difference learning | |
| CN116844696A (en) | Facial image traditional Chinese medicine constitution identification method based on deep neural network | |
| CN112488936B (en) | Method for deblurring finger vein blurred image based on generation countermeasure network | |
| CN117611548A (en) | Image quality evaluation method and system based on distortion information | |
| CN115527253A (en) | Attention mechanism-based lightweight facial expression recognition method and system | |
| CN117456339B (en) | Image quality evaluation method and system based on multi-level feature multiplexing |
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 |