WO2024078394A1 - Image quality evaluation method and apparatus, and electronic device, storage medium and program product - Google Patents
Image quality evaluation method and apparatus, and electronic device, storage medium and program product Download PDFInfo
- Publication number
- WO2024078394A1 WO2024078394A1 PCT/CN2023/123283 CN2023123283W WO2024078394A1 WO 2024078394 A1 WO2024078394 A1 WO 2024078394A1 CN 2023123283 W CN2023123283 W CN 2023123283W WO 2024078394 A1 WO2024078394 A1 WO 2024078394A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- quantum
- data
- image
- image data
- evaluated
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 59
- 238000013441 quality evaluation Methods 0.000 title claims abstract description 17
- 238000013528 artificial neural network Methods 0.000 claims abstract description 73
- 238000011156 evaluation Methods 0.000 claims abstract description 50
- 230000006870 function Effects 0.000 claims description 58
- 238000005259 measurement Methods 0.000 claims description 45
- 230000015654 memory Effects 0.000 claims description 31
- 238000010276 construction Methods 0.000 claims description 15
- 238000000691 measurement method Methods 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 13
- 238000013507 mapping Methods 0.000 claims description 12
- 238000001303 quality assessment method Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 description 15
- 238000010586 diagram Methods 0.000 description 8
- 230000001360 synchronised effect Effects 0.000 description 5
- 238000004891 communication Methods 0.000 description 4
- 238000012216 screening Methods 0.000 description 4
- 101000827703 Homo sapiens Polyphosphoinositide phosphatase Proteins 0.000 description 3
- 102100023591 Polyphosphoinositide phosphatase Human genes 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000007430 reference method Methods 0.000 description 3
- 101100012902 Saccharomyces cerevisiae (strain ATCC 204508 / S288c) FIG2 gene Proteins 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 101001121408 Homo sapiens L-amino-acid oxidase Proteins 0.000 description 1
- 102100026388 L-amino-acid oxidase Human genes 0.000 description 1
- 101100233916 Saccharomyces cerevisiae (strain ATCC 204508 / S288c) KAR5 gene Proteins 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000002096 quantum dot Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/98—Detection or correction of errors, e.g. by rescanning the pattern or by human intervention; Evaluation of the quality of the acquired patterns
- G06V10/993—Evaluation of the quality of the acquired pattern
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N10/00—Quantum computing, i.e. information processing based on quantum-mechanical phenomena
- G06N10/70—Quantum error correction, detection or prevention, e.g. surface codes or magic state distillation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
- G06V10/765—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects using rules for classification or partitioning the feature space
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Definitions
- the embodiments of the present disclosure relate to the field of computer technology, and in particular, to an image quality evaluation method, device, electronic device, storage medium, and program product.
- digital image information has the advantages of rich information and intuitive and easy to understand, and is an important source of information for people to acquire knowledge.
- people's accumulation of digital image processing technology has continued to mature, and image transmission has become more frequent.
- information loss will occur in the process of transmitting or storing various images, and image quality will decrease. How to select higher quality images has become a difficult problem.
- the embodiments of the present disclosure provide an image quality evaluation method, device, electronic device, storage medium and program product.
- an embodiment of the present disclosure provides an image quality evaluation method, the method comprising:
- the subjective evaluation value corresponding to the image to be evaluated is obtained based on an image quality score of the image to be evaluated by at least one image observer.
- the constructing of a quantum classifier includes:
- the initial quantum neural network is trained using the first quantum data corresponding to each image data to obtain a quantum classifier.
- the using the first quantum data corresponding to each image data to train an initial quantum neural network to obtain a quantum classifier includes:
- a loss function of the initial quantum neural network is determined based on the second measurement data and the first quantum data.
- the using the first quantum data corresponding to each image data to train an initial quantum neural network to obtain a quantum classifier further includes:
- the parameters of the initial quantum neural network corresponding to the loss function are determined as target parameters, and the quantum classifier is obtained based on the target parameters.
- each image data in the image data set to image data in a preset form; determining quantum state data of an initial state based on the image data in the preset form corresponding to each image data;
- each image data is encoded into a quantum gate form and acts Based on the quantum state data of the initial state, first quantum data corresponding to each image data is obtained.
- the using the quantum classifier to determine a target image whose image quality satisfies a first condition among the multiple images to be evaluated includes:
- the decision boundary generated by the quantum classifier is used to determine the area where each image to be evaluated among the multiple images to be evaluated is located in the decision boundary, and the image to be evaluated corresponding to the subjective evaluation value located in the central area of the decision boundary is determined as the target image whose image quality meets the first condition among the multiple images to be evaluated.
- an embodiment of the present disclosure provides an image quality assessment device, the device comprising:
- a building unit configured to build a quantum classifier
- a determination unit is configured to input subjective evaluation values corresponding to a plurality of images to be evaluated into the quantum classifier, and use the quantum classifier to determine a target image whose image quality meets a first condition among the plurality of images to be evaluated;
- the subjective evaluation value corresponding to the image to be evaluated is obtained based on an image quality score of the image to be evaluated by at least one image observer.
- the construction unit is configured to obtain an image data set, encode each image data included in the image data set into corresponding first quantum data; and train an initial quantum neural network using the first quantum data corresponding to each image data to obtain a quantum classifier.
- the construction unit is configured to input the first quantum data corresponding to each image data into an initial quantum neural network to obtain second quantum data; measure the second quantum data using a preset measurement method to obtain first measurement data corresponding to the second quantum data;
- the first measurement data is biased to obtain second measurement data located in a preset mapping interval; the loss function of the initial quantum neural network is determined based on the second measurement data and the first quantum data; it is determined whether the loss function satisfies a second condition; if the loss function does not satisfy the second condition, the following steps are repeated until the loss function satisfies the second condition: the first quantum data corresponding to each image data is input into the initial quantum neural network to obtain second quantum data; the second quantum data is measured using a preset measurement method to obtain first measurement data corresponding to the second quantum data; the first measurement data is biased to obtain second measurement data located in a preset mapping interval; the loss function of the initial quantum neural network is determined based on the second measurement data and the first quantum data.
- the construction unit is configured to, when the loss function satisfies the second condition, determine the parameters of the initial quantum neural network corresponding to the loss function as target parameters, and obtain the quantum classifier based on the target parameters.
- each image data in the image data set is N-dimensional data, where N is an integer greater than or equal to 1; the construction unit is configured to set each image data in the image data set as image data in a preset form; determine quantum state data of an initial state based on the image data in the preset form corresponding to each image data; based on the image data in the preset form corresponding to each image data, encode each image data into a quantum gate form and act on the quantum state data of the initial state to obtain first quantum data corresponding to each image data.
- the determination unit is configured to use the decision boundary generated by the quantum classifier to determine the area where each image to be evaluated among the multiple images to be evaluated is located in the decision boundary, and determine the image to be evaluated corresponding to the subjective evaluation value located in the central area of the decision boundary as the target image whose image quality meets the first condition among the multiple images to be evaluated.
- an embodiment of the present disclosure provides an electronic device, comprising: a memory and a processor, wherein the memory stores computer executable instructions, and when the processor runs the computer executable instructions on the memory, the image quality evaluation method described in the above embodiment can be implemented.
- an embodiment of the present disclosure provides a computer storage medium having executable instructions stored thereon, which, when executed by a processor, implements the image quality evaluation method described in the above embodiment.
- an embodiment of the present disclosure provides a computer program product, including a computer-readable code.
- a processor in the electronic device executes the above-mentioned image quality assessment method.
- the technical solution of the embodiment of the present disclosure constructs a quantum classifier; inputs the subjective evaluation values corresponding to multiple images to be evaluated into the quantum classifier, and uses the quantum classifier to determine the target image whose image quality meets the first condition among the multiple images to be evaluated.
- the quantum classifier generated by the quantum neural network can be used to classify the subjective evaluation values of the images and select images with better quality from multiple images.
- FIG1 is a schematic diagram of a flow chart of an image quality evaluation method provided by an embodiment of the present disclosure
- FIG2 is an image of different qualities provided by an embodiment of the present disclosure.
- FIG3 is a schematic diagram of a construction process of a quantum classifier provided in an embodiment of the present disclosure
- FIG4 is a schematic diagram of the structure of an image quality assessment device provided by an embodiment of the present disclosure.
- FIG5 is a schematic diagram of the structural composition of an electronic device provided in an embodiment of the present disclosure.
- a and/or B may indicate the existence of A alone, A and B at the same time, and B alone.
- at least one herein indicates any combination of at least two of any one or more of a plurality of.
- at least one of A, B, and C may indicate any one or more elements selected from the set consisting of A, B, and C.
- Image quality evaluation methods can be divided into subjective evaluation methods and objective evaluation methods. The following introduces the subjective evaluation method and the objective evaluation method.
- the subjective evaluation method prepares a set of image samples in advance, and then evaluates the image quality by observing the images to obtain the subjective evaluation results.
- the subjective evaluation results are then processed and the subjective evaluation value, mean and standard deviation in a group of images are set as the basis for judging the image quality.
- the semi-reference method combines machine learning methods to evaluate image quality. It assumes that images of similar quality have the same rules in underlying features. It does not analyze the causes of image distortion and the methods of designing features. It simply uses the features obtained by computer learning as the evaluation criteria for the image.
- image quality assessment methods based on machine learning is how to evaluate image quality without being disturbed by image content, especially when extracting specific feature statistics from images, the image content will bring a large deviation to the feature statistics; or when extracting features from images with very complex content, it will bring more feature statistics than images of the same quality with simple content.
- image quality assessment methods since there are no actual images as references, in order to balance the evaluation of multiple categories of images, it is easy to use the same method to "over-estimate" complex images and "under-estimate” simple images, resulting in problems in making consistent evaluations of all samples.
- the technical solution of the embodiment of the present disclosure uses a quantum neural network method to classify images.
- the designed quantum neural network can select better images from images obtained through subjective evaluation. Compared with the method of simply using subjective evaluation values to determine image quality, the technical solution of the embodiment of the present disclosure is conducive to faster and better selection of better images.
- the technical solution of the disclosed embodiment can use quantum coding to convert traditional classical image data into quantum data coding, and then obtain classification system parameters by training quantum neural network, and obtain quantum classifier with better image quality screening effect by screening better classification system parameters. After that, the subjective evaluation value of the image is used as the input data of the quantum classifier, and then the image quality classification work is carried out. According to the set threshold, the candidate images with better subjective quality can be automatically screened.
- the image quality evaluation method of the disclosed embodiment is introduced as follows.
- FIG. 1 is a flow chart of an image quality evaluation method provided by an embodiment of the present disclosure.
- the image quality evaluation method of the embodiment of the present disclosure includes the following steps:
- Step 101 Build a quantum classifier.
- the quantum classifier is a quantum classifier for evaluating image quality obtained by training an initial quantum neural network.
- step 101 includes the following steps:
- the data that the quantum neural network needs to process is quantum data
- the traditional digital image data needs to be converted into data in quantum form before the subsequent training of the initial quantum neural network can be performed to obtain a quantum classifier for evaluating image quality.
- each image data in the image data set is N-dimensional data, N is an integer greater than or equal to 1, and the above step 1.1) includes the following steps:
- each image data is encoded into a quantum gate format and acts on the quantum state data of the initial state to obtain the first quantum data corresponding to each image data.
- the two-dimensional image data needs to be encoded first to obtain quantum data corresponding to each two-dimensional image data.
- the encoding process is as follows:
- the quantum state of the initial state can be set to
- the two-dimensional image data ⁇ x k ⁇ can be encoded into the form of a quantum gate U(x k ) and act on the initial quantum state to obtain a series of quantum data as shown in the following formula (1).
- This expression process is the encoding process of converting classical image data information into quantum digital information.
- ⁇ > k U(x k )
- the quantum gate can be set to the form of the following formula (2):
- the quantum gate constructed with 2 bits can be expressed in the form of the following formula (3):
- step 1.2) includes the following steps:
- the first quantum data obtained through step 1.1) can also be called quantum input state data, that is, the quantum data to be input into the initial quantum neural network. After the first quantum data is input into the initial quantum neural network, quantum output state data is obtained.
- the entire initial quantum neural network is set to U( ⁇ ), and the matrix form of the following formula (5) is used as the intermediate revolving gate:
- the 2-bit quantum gate matrix obtained by the above formula (4) can be expressed in the form of the following formula (6):
- the quantum state is input into the initial quantum neural network.
- the quantum output state calculated by the initial quantum neural network is
- ⁇ > U( ⁇ )
- the first of the above-defined quantum bit sequences can be used as the expected value by measuring the Pauli Z operator, which can be defined as
- the preset measurement method in the embodiments of the present disclosure includes but is not limited to the Pauli Z operator measurement method, and can also be any measurement method that can measure the second quantum data, which is not limited in the embodiments of the present disclosure.
- the value range of ⁇ Z> after measurement can be specified as [-1, 1], and correspondingly,
- the square loss function is used as the mapping interval of the loss function when training the initial quantum neural network, so that the range of the loss function during the quantum neural network training process is between [0,1].
- determining whether the loss function satisfies the second condition may be determining whether the loss function is less than a set value, such as 0.001. In the case where the loss function is less than the set value, the quantum neural network is considered to have been trained. In another embodiment, determining whether the loss function satisfies the second condition may be determining whether the loss function is a minimized loss function. In the case where the loss function is a minimized loss function, the quantum neural network is considered to have been trained.
- the loss function does not satisfy the second condition, it is necessary to update the training parameters of the quantum neural network and repeatedly train the quantum neural network using the first quantum data until the loss function of the trained neural network satisfies the second condition.
- step 1.2) further includes the following steps:
- the parameters of the initial quantum neural network corresponding to the loss function are determined as target parameters, and the quantum classifier is obtained based on the target parameters.
- the quantum neural network When the loss function satisfies the second condition, the quantum neural network is considered to have been trained. At this time, the network parameters corresponding to the trained quantum neural network can be obtained.
- the quantum neural network obtained by taking the network parameters is the quantum classifier used to evaluate the image quality.
- the obtained quantum classifier can be used to evaluate the image quality of a group of images to be evaluated.
- Step 102 Inputting subjective evaluation values corresponding to a plurality of images to be evaluated into the quantum classifier, and using the quantum classifier to determine a target image whose image quality meets a first condition among the plurality of images to be evaluated.
- the subjective evaluation value corresponding to the image to be evaluated is obtained based on the image quality score of the image to be evaluated by at least one image observer.
- the image quality can be judged by subjective scoring by image observers.
- the method that can be adopted is: let the image observer make a quality judgment on the image to be evaluated according to the visual effect based on the evaluation criteria specified in advance or his own subjective experience, and give a quality score, that is, give an image quality score, and perform a weighted average of the quality scores given by all image observers.
- the result obtained is the subjective evaluation value of the image, which can also be called a subjective scoring value. The higher the subjective evaluation value, the better the image quality.
- the image whose image quality satisfies the first condition among the multiple images to be evaluated may be an image with better image quality among the multiple images to be evaluated, that is, an image with a higher image quality ranking.
- the obtained quantum classifier can be used to evaluate the image quality of a group of images to be evaluated.
- the subjective evaluation values corresponding to a group of multiple images to be evaluated can be input into the trained quantum classifier, and the quantum classifier can be used to screen out images with better image quality from the group of images.
- step 102 may be implemented by the following steps:
- the decision boundary generated by the quantum classifier is used to determine the area where each image to be evaluated among the multiple images to be evaluated is located in the decision boundary, and the image to be evaluated corresponding to the subjective evaluation value located in the central area of the decision boundary is determined as the target image whose image quality meets the first condition among the multiple images to be evaluated.
- a subjective image quality evaluation value dataset can be selected, and the image quality subjective evaluation value dataset can be input into the constructed quantum classifier using the quantum classifier constructed in step 101, and the decision boundary generated by the quantum classifier can be used to determine the subjective image quality evaluation value dataset. Different image regions are distinguished, and the image corresponding to the dmos value (i.e., subjective evaluation value) in the center of the quantum decision boundary is the image with better quality. Table 1 below is a set of image quality scoring data sets.
- image represents the image number
- dst_idx represents the image encoding method
- dst_lev represents the image level
- dmos_std represents the mean subjective score difference
- dmos represents the subjective evaluation value.
- FIG2 shows images of different qualities provided by an embodiment of the present disclosure.
- the five images (a) to (e) in the figure have different main The subjective evaluation values of the five images in Figure 2 are input into the trained quantum classifier. According to the different subjective evaluation values of each image, the leftmost image (a) in Figure 2 can be selected as the image with the best visual quality. In Figure 2, the image quality of the five images decreases from left to right.
- the present invention can utilize a quantum classifier generated by a quantum neural network to classify subjective evaluation values of images and select images with better quality from multiple images.
- the image quality evaluation method provided by the embodiment of the present disclosure is different from the traditional machine learning method for generating a classifier.
- the embodiment of the present disclosure uses a quantum neural network method to generate a classifier, which is different from the basic principle of the traditional image screening method.
- the technical solution of the embodiment of the present disclosure uses a classifier generated by a quantum neural network to classify the subjective evaluation values of the image and screen out images with better quality, which has good application value.
- the technical solution of the embodiment of the present disclosure is different from the traditional method running in a classical computer.
- the solution of the embodiment of the present disclosure can be adapted to run in electronic computers and quantum computer modes, and has a high value of frontier technology pre-research for computers using quantum modes in the future.
- Figure 3 is a schematic diagram of the construction process of the quantum classifier provided by the embodiment of the present disclosure.
- Figure 3 involves the processing process of image data and the training process of the quantum neural network.
- the generation process of the quantum classifier is divided into two parts: quantum data encoding and quantum neural network training.
- the quantum data encoding part converts the classical digital image information into quantum data information;
- the process of obtaining the classifier by quantum neural network training is to train the initial quantum neural network based on the obtained quantum data information to obtain the classification system parameters of the quantum classifier, thereby obtaining the quantum classifier.
- the process of quantum data encoding and the process of obtaining the quantum classifier by quantum neural network training can refer to the above step 101.
- FIG. 4 is a schematic diagram of the structure of an image quality assessment device provided by an embodiment of the present disclosure. As shown in FIG. 4 , the device includes:
- a construction unit 401 is configured to construct a quantum classifier
- a determination unit 402 is configured to input subjective evaluation values corresponding to a plurality of images to be evaluated into the quantum classifier, and use the quantum classifier to determine a target image whose image quality meets a first condition among the plurality of images to be evaluated;
- the subjective evaluation value corresponding to the image to be evaluated is obtained based on an image quality score of the image to be evaluated by at least one image observer.
- the construction unit 401 is configured to obtain an image data set, encode each image data included in the image data set into corresponding first quantum data; and train an initial quantum neural network using the first quantum data corresponding to each image data to obtain a quantum classifier.
- the construction unit 401 is configured to input the first quantum data corresponding to each image data into the initial quantum neural network to obtain second quantum data; measure the second quantum data using a preset measurement method to obtain first measurement data corresponding to the second quantum data; bias the first measurement data to obtain second measurement data located in a preset mapping interval; determine the loss function of the initial quantum neural network based on the second measurement data and the first quantum data; determine whether the loss function satisfies a second condition; if the loss function does not satisfy the second condition, repeat the following steps until the loss function satisfies the second condition: input the first quantum data corresponding to each image data into the initial quantum neural network to obtain second quantum data; measure the second quantum data using a preset measurement method to obtain first measurement data corresponding to the second quantum data; bias the first measurement data to obtain second measurement data located in a preset mapping interval; determine the loss function of the initial quantum neural network based on the second measurement data and the first quantum data.
- the construction unit 401 is configured to, when the loss function satisfies the second condition, determine the parameters of the initial quantum neural network corresponding to the loss function as target parameters, and obtain the quantum classifier based on the target parameters.
- each image data in the image data set is N-dimensional data, where N is an integer greater than or equal to 1; the construction unit 401 is configured to set each image data in the image data set as image data in a preset form; determine quantum state data of an initial state based on the image data in the preset form corresponding to each image data; based on the image data in the preset form corresponding to each image data, encode each image data into a quantum gate form and act on the quantum state data of the initial state to obtain first quantum data corresponding to each image data.
- the determination unit 402 is configured to use the decision boundary generated by the quantum classifier to determine the area where each image to be evaluated among the multiple images to be evaluated is located in the decision boundary, and determine the image to be evaluated corresponding to the subjective evaluation value located in the central area of the decision boundary as the target image whose image quality meets the first condition among the multiple images to be evaluated.
- each unit in the image quality assessment device shown in FIG4 can be understood by referring to the relevant description of the above-mentioned image quality assessment method.
- the functions of each unit in the image quality assessment device shown in FIG4 can be implemented by a program running on a processor, or by a logic circuit.
- FIG5 is a schematic diagram of the hardware structure of the electronic device of the embodiment of the present disclosure.
- the electronic device includes: a communication component 503 configured to perform data transmission, at least one processor 501, and a memory 502 configured to store a computer program that can be run on the processor 501.
- the various components in the terminal are coupled together through a bus system 504. It can be understood that the bus system 504 is configured to achieve connection and communication between these components.
- the bus system 504 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, various buses are marked as bus systems 504 in FIG5.
- the processor 501 executes the computer program, it at least performs the steps of the method shown in FIG. 1 .
- the memory 502 can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories.
- the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM); the magnetic surface memory can be a disk memory or a tape memory.
- the volatile memory can be a random access memory (RAM), which is used as an external cache.
- RAM random access memory
- SRAM static random access memory
- SSRAM synchronous static random access memory
- DRAM dynamic random access memory
- SDRAM synchronous dynamic random access memory
- DDRSDRAM double data rate synchronous dynamic random access memory
- ESDRAM enhanced synchronous dynamic random access memory
- SLDRAM synchronous link dynamic random access memory
- DRRAM direct memory bus random access memory
- the method disclosed in the above embodiment of the present disclosure can be applied to the processor 501, or implemented by the processor 501.
- the processor 501 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit in the processor 501 or the instruction in the form of software.
- the above processor 501 can be a general-purpose processor, a DSP, or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc.
- the processor 501 can implement or execute the methods, steps and logic block diagrams disclosed in the embodiment of the present disclosure.
- the general-purpose processor can be a microprocessor or any conventional processor, etc.
- the steps of the method disclosed in the embodiment of the present disclosure can be directly embodied as a hardware decoding processor to execute, or the hardware and software modules in the decoding processor can be combined to execute.
- the software module can be located in a storage medium, which is located in the memory 502.
- the processor 501 reads the information in the memory 502 and completes the steps of the above method in combination with its hardware.
- the electronic device can be implemented by one or more application specific integrated circuits (ASIC), DSP, programmable logic device (PLD), complex programmable logic device (CPLD), FPGA, general-purpose processor, controller, MCU, microprocessor, or other electronic components to execute the aforementioned image quality evaluation method.
- ASIC application specific integrated circuits
- DSP digital signal processor
- PLD programmable logic device
- CPLD complex programmable logic device
- FPGA field-programmable gate array
- controller microprocessor
- microprocessor microprocessor
- the embodiment of the present disclosure further provides a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, it is used to at least perform the steps of the method shown in Figure 1.
- the computer-readable storage medium may be a memory.
- the memory may be the memory 502 shown in Figure 5.
- An embodiment of the present disclosure provides a computer program product, including a computer-readable code.
- a processor in the electronic device executes the above-mentioned image quality assessment method.
- the disclosed methods and intelligent devices can be implemented in other ways.
- the device embodiments described above are schematic.
- the division of the units is a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed.
- the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.
- the units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.
- all functional units in the embodiments of the present disclosure may be integrated into a second processing unit, or each unit may be separately configured as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.
- the quantum classifier generated by the quantum neural network can be used to classify the subjective evaluation values of the images, thereby screening out images with better quality from the plurality of images.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Quality & Reliability (AREA)
- Condensed Matter Physics & Semiconductors (AREA)
- Computational Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
- Image Analysis (AREA)
Abstract
Disclosed in the present disclosure are an image quality evaluation method and apparatus, and an electronic device and a storage medium. The method comprises: constructing a quantum classifier; and inputting, into the quantum classifier, subjective evaluation values corresponding to a plurality of images to be evaluated, and determining, by using the quantum classifier and from among the plurality of images to be evaluated, a target image, the image quality of which meets a first condition. The present disclosure can make it possible to classify subjective evaluation values of images by using a quantum classifier, which is generated by means of a quantum neural network, so as to pick out an image with a better quality from among a plurality of images.
Description
相关申请的交叉引用CROSS-REFERENCE TO RELATED APPLICATIONS
本公开基于申请号为202211252465.7,申请日为2022年10月13日、申请名称为“一种图像质量评价方法、装置、电子设备及存储介质”的中国专利申请提出,并要求该中国专利申请的优先权,该中国专利申请的全部内容在此以全文引入的方式引入本公开。This disclosure is based on a Chinese patent application with application number 202211252465.7, application date October 13, 2022, and application name “A method, device, electronic device and storage medium for evaluating image quality”, and claims the priority of the Chinese patent application. The entire contents of the Chinese patent application are hereby introduced into this disclosure in their entirety.
本公开实施例涉及计算机技术领域,尤其涉及一种图像质量评价方法、装置、电子设备、存储介质及程序产品。The embodiments of the present disclosure relate to the field of computer technology, and in particular, to an image quality evaluation method, device, electronic device, storage medium, and program product.
数字图像信息作为从客观世界获取数据信息的方法之一,具有信息量丰富且直观易懂的优点,是人们获取知识的重要信息来源。随着计算机系统的快速发展,人们对数字图像处理技术的积累不断成熟,图像的传输更为频繁,随之而来的是,在对各种图像进行传输或存储的过程中会出现信息的丢失,图像质量出现下降,如何进行较优质量图像的选取成为一个难题。As one of the methods to obtain data information from the objective world, digital image information has the advantages of rich information and intuitive and easy to understand, and is an important source of information for people to acquire knowledge. With the rapid development of computer systems, people's accumulation of digital image processing technology has continued to mature, and image transmission has become more frequent. As a result, information loss will occur in the process of transmitting or storing various images, and image quality will decrease. How to select higher quality images has become a difficult problem.
发明内容Summary of the invention
为解决上述技术问题,本公开实施例提供了一种图像质量评价方法、装置、电子设备、存储介质及程序产品。To solve the above technical problems, the embodiments of the present disclosure provide an image quality evaluation method, device, electronic device, storage medium and program product.
第一方面,本公开实施例提供了一种图像质量评价方法,所述方法包括:In a first aspect, an embodiment of the present disclosure provides an image quality evaluation method, the method comprising:
构建量子分类器;Building a quantum classifier;
将多个待评价图像对应的主观评价值输入所述量子分类器,利用所述量子分类器确定所述多个待评价图像中图像质量满足第一条件的目标图像;Inputting subjective evaluation values corresponding to a plurality of images to be evaluated into the quantum classifier, and using the quantum classifier to determine a target image among the plurality of images to be evaluated whose image quality meets a first condition;
其中,针对所述多个待评价图像中的每个待评价图像,该待评价图像对应的主观评价值基于至少一个图像观测者对该待评价图像的图像质量评分得到。For each image to be evaluated among the multiple images to be evaluated, the subjective evaluation value corresponding to the image to be evaluated is obtained based on an image quality score of the image to be evaluated by at least one image observer.
在一种可能的实现方式中,所述构建量子分类器,包括:In a possible implementation, the constructing of a quantum classifier includes:
获取图像数据集,将所述图像数据集中包括的各图像数据编码为对应的第一量子数据;Acquire an image data set, and encode each image data included in the image data set into corresponding first quantum data;
利用所述各图像数据对应的第一量子数据训练初始量子神经网络,得到量子分类器。The initial quantum neural network is trained using the first quantum data corresponding to each image data to obtain a quantum classifier.
在一种可能的实现方式中,所述利用所述各图像数据对应的第一量子数据训练初始量子神经网络,得到量子分类器,包括:In a possible implementation, the using the first quantum data corresponding to each image data to train an initial quantum neural network to obtain a quantum classifier includes:
将所述各图像数据对应的第一量子数据输入至初始量子神经网络,得到第二量子数据;Inputting the first quantum data corresponding to each image data into the initial quantum neural network to obtain second quantum data;
利用预设测量方法测量所述第二量子数据,得到所述第二量子数据对应的第一测量数据;Measuring the second quantum data using a preset measurement method to obtain first measurement data corresponding to the second quantum data;
对所述第一测量数据进行偏置处理得到位于预设映射区间的第二测量数据;Performing offset processing on the first measurement data to obtain second measurement data located in a preset mapping interval;
基于所述第二测量数据以及所述第一量子数据确定所述初始量子神经网络的损失函数;Determine a loss function of the initial quantum neural network based on the second measurement data and the first quantum data;
确定所述损失函数是否满足第二条件;在所述损失函数不满足所述第二条件的情况下,重复执行如下步骤,直至所述损失函数满足所述第二条件:Determine whether the loss function satisfies a second condition; if the loss function does not satisfy the second condition, repeat the following steps until the loss function satisfies the second condition:
将所述各图像数据对应的第一量子数据输入至初始量子神经网络,得到第二量子数据;Inputting the first quantum data corresponding to each image data into the initial quantum neural network to obtain second quantum data;
利用预设测量方法测量所述第二量子数据,得到所述第二量子数据对应的第一测量数据;Measuring the second quantum data using a preset measurement method to obtain first measurement data corresponding to the second quantum data;
对所述第一测量数据进行偏置处理得到位于预设映射区间的第二测量数据;Performing offset processing on the first measurement data to obtain second measurement data located in a preset mapping interval;
基于所述第二测量数据以及所述第一量子数据确定所述初始量子神经网络的损失函数。A loss function of the initial quantum neural network is determined based on the second measurement data and the first quantum data.
在一种可能的实现方式中,所述利用所述各图像数据对应的第一量子数据训练初始量子神经网络,得到量子分类器,还包括:In a possible implementation, the using the first quantum data corresponding to each image data to train an initial quantum neural network to obtain a quantum classifier further includes:
在所述损失函数满足第二条件的情况下,将所述损失函数对应的所述初始量子神经网络的参数确定为目标参数,并基于所述目标参数得到所述量子分类器。When the loss function satisfies the second condition, the parameters of the initial quantum neural network corresponding to the loss function are determined as target parameters, and the quantum classifier is obtained based on the target parameters.
在一种可能的实现方式中,所述图像数据集中的各图像数据为N维数据,N为大于等于1的整数;所述将所述图像数据集中包括的各图像数据编码为对应的第一量子数据,包括:In a possible implementation manner, each image data in the image data set is N-dimensional data, where N is an integer greater than or equal to 1; encoding each image data included in the image data set into corresponding first quantum data includes:
将所述图像数据集中的各图像数据设置为预设形式的图像数据;基于所述各图像数据对应的预设形式的图像数据确定初始状态的量子态数据;Setting each image data in the image data set to image data in a preset form; determining quantum state data of an initial state based on the image data in the preset form corresponding to each image data;
基于所述各图像数据对应的预设形式的图像数据,将所述各图像数据编码为量子门形式并作用
在所述初始状态的量子态数据上,得到所述各图像数据对应的第一量子数据。Based on the image data in a preset form corresponding to each image data, each image data is encoded into a quantum gate form and acts Based on the quantum state data of the initial state, first quantum data corresponding to each image data is obtained.
在一种可能的实现方式中,所述利用所述量子分类器确定所述多个待评价图像中图像质量满足第一条件的目标图像,包括:In a possible implementation, the using the quantum classifier to determine a target image whose image quality satisfies a first condition among the multiple images to be evaluated includes:
利用所述量子分类器产生的决策边界确定所述多个待评价图像中的各待评价图像在所述决策边界中所处的区域,将位于所述决策边界的中心区域的主观评价值对应的待评价图像确定为所述多个待评价图像中图像质量满足第一条件的目标图像。The decision boundary generated by the quantum classifier is used to determine the area where each image to be evaluated among the multiple images to be evaluated is located in the decision boundary, and the image to be evaluated corresponding to the subjective evaluation value located in the central area of the decision boundary is determined as the target image whose image quality meets the first condition among the multiple images to be evaluated.
第二方面,本公开实施例提供了一种图像质量评价装置,所述装置包括:In a second aspect, an embodiment of the present disclosure provides an image quality assessment device, the device comprising:
构建单元,配置为构建量子分类器;a building unit configured to build a quantum classifier;
确定单元,配置为将多个待评价图像对应的主观评价值输入所述量子分类器,利用所述量子分类器确定所述多个待评价图像中图像质量满足第一条件的目标图像;A determination unit is configured to input subjective evaluation values corresponding to a plurality of images to be evaluated into the quantum classifier, and use the quantum classifier to determine a target image whose image quality meets a first condition among the plurality of images to be evaluated;
其中,针对所述多个待评价图像中的每个待评价图像,该待评价图像对应的主观评价值基于至少一个图像观测者对该待评价图像的图像质量评分得到。For each image to be evaluated among the multiple images to be evaluated, the subjective evaluation value corresponding to the image to be evaluated is obtained based on an image quality score of the image to be evaluated by at least one image observer.
在一种可能的实现方式中,所述构建单元,配置为获取图像数据集,将所述图像数据集中包括的各图像数据编码为对应的第一量子数据;利用所述各图像数据对应的第一量子数据训练初始量子神经网络,得到量子分类器。In a possible implementation, the construction unit is configured to obtain an image data set, encode each image data included in the image data set into corresponding first quantum data; and train an initial quantum neural network using the first quantum data corresponding to each image data to obtain a quantum classifier.
在一种可能的实现方式中,所述构建单元,配置为将所述各图像数据对应的第一量子数据输入至初始量子神经网络,得到第二量子数据;利用预设测量方法测量所述第二量子数据,得到所述第二量子数据对应的第一测量数据;In a possible implementation, the construction unit is configured to input the first quantum data corresponding to each image data into an initial quantum neural network to obtain second quantum data; measure the second quantum data using a preset measurement method to obtain first measurement data corresponding to the second quantum data;
对所述第一测量数据进行偏置处理得到位于预设映射区间的第二测量数据;基于所述第二测量数据以及所述第一量子数据确定所述初始量子神经网络的损失函数;确定所述损失函数是否满足第二条件;在所述损失函数不满足所述第二条件的情况下,重复执行如下步骤,直至所述损失函数满足所述第二条件:将所述各图像数据对应的第一量子数据输入至初始量子神经网络,得到第二量子数据;利用预设测量方法测量所述第二量子数据,得到所述第二量子数据对应的第一测量数据;对所述第一测量数据进行偏置处理得到位于预设映射区间的第二测量数据;基于所述第二测量数据以及所述第一量子数据确定所述初始量子神经网络的损失函数。The first measurement data is biased to obtain second measurement data located in a preset mapping interval; the loss function of the initial quantum neural network is determined based on the second measurement data and the first quantum data; it is determined whether the loss function satisfies a second condition; if the loss function does not satisfy the second condition, the following steps are repeated until the loss function satisfies the second condition: the first quantum data corresponding to each image data is input into the initial quantum neural network to obtain second quantum data; the second quantum data is measured using a preset measurement method to obtain first measurement data corresponding to the second quantum data; the first measurement data is biased to obtain second measurement data located in a preset mapping interval; the loss function of the initial quantum neural network is determined based on the second measurement data and the first quantum data.
在一种可能的实现方式中,所述构建单元,配置为在所述损失函数满足第二条件的情况下,将所述损失函数对应的所述初始量子神经网络的参数确定为目标参数,并基于所述目标参数得到所述量子分类器。In a possible implementation manner, the construction unit is configured to, when the loss function satisfies the second condition, determine the parameters of the initial quantum neural network corresponding to the loss function as target parameters, and obtain the quantum classifier based on the target parameters.
在一种可能的实现方式中,所述图像数据集中的各图像数据为N维数据,N为大于等于1的整数;所述构建单元,配置为将所述图像数据集中的各图像数据设置为预设形式的图像数据;基于所述各图像数据对应的预设形式的图像数据确定初始状态的量子态数据;基于所述各图像数据对应的预设形式的图像数据,将所述各图像数据编码为量子门形式并作用在所述初始状态的量子态数据上,得到所述各图像数据对应的第一量子数据。In a possible implementation, each image data in the image data set is N-dimensional data, where N is an integer greater than or equal to 1; the construction unit is configured to set each image data in the image data set as image data in a preset form; determine quantum state data of an initial state based on the image data in the preset form corresponding to each image data; based on the image data in the preset form corresponding to each image data, encode each image data into a quantum gate form and act on the quantum state data of the initial state to obtain first quantum data corresponding to each image data.
在一种可能的实现方式中,所述确定单元,配置为利用所述量子分类器产生的决策边界确定所述多个待评价图像中的各待评价图像在所述决策边界中所处的区域,将位于所述决策边界的中心区域的主观评价值对应的待评价图像确定为所述多个待评价图像中图像质量满足第一条件的目标图像。In a possible implementation, the determination unit is configured to use the decision boundary generated by the quantum classifier to determine the area where each image to be evaluated among the multiple images to be evaluated is located in the decision boundary, and determine the image to be evaluated corresponding to the subjective evaluation value located in the central area of the decision boundary as the target image whose image quality meets the first condition among the multiple images to be evaluated.
第三方面,本公开实施例提供了一种电子设备,所述电子设备包括:存储器和处理器,所述存储器上存储有计算机可执行指令,所述处理器运行所述存储器上的计算机可执行指令时可实现上述实施例所述的图像质量的评价方法。In a third aspect, an embodiment of the present disclosure provides an electronic device, comprising: a memory and a processor, wherein the memory stores computer executable instructions, and when the processor runs the computer executable instructions on the memory, the image quality evaluation method described in the above embodiment can be implemented.
第四方面,本公开实施例提供了一种计算机存储介质,所述存储介质上存储有可执行指令,该可执行指令被处理器执行时实现上述实施例所述的图像质量的评价方法。In a fourth aspect, an embodiment of the present disclosure provides a computer storage medium having executable instructions stored thereon, which, when executed by a processor, implements the image quality evaluation method described in the above embodiment.
第五方面,本公开实施例提供了一种计算机程序产品,包括计算机可读代码,在所述计算机可读代码在电子设备中运行的情况下,所述电子设备中的处理器执行上述的图像质量评价方法。In a fifth aspect, an embodiment of the present disclosure provides a computer program product, including a computer-readable code. When the computer-readable code runs in an electronic device, a processor in the electronic device executes the above-mentioned image quality assessment method.
本公开实施例的技术方案,通过构建量子分类器;将多个待评价图像对应的主观评价值输入所述量子分类器,利用所述量子分类器确定所述多个待评价图像中图像质量满足第一条件的目标图像。如此,能够利用量子神经网络生成的量子分类器通过对图像主观评价值进行分类,从多幅图像中筛选出质量较优的图像。The technical solution of the embodiment of the present disclosure constructs a quantum classifier; inputs the subjective evaluation values corresponding to multiple images to be evaluated into the quantum classifier, and uses the quantum classifier to determine the target image whose image quality meets the first condition among the multiple images to be evaluated. In this way, the quantum classifier generated by the quantum neural network can be used to classify the subjective evaluation values of the images and select images with better quality from multiple images.
图1为本公开实施例提供的图像质量评价方法的流程示意图;FIG1 is a schematic diagram of a flow chart of an image quality evaluation method provided by an embodiment of the present disclosure;
图2为本公开实施例提供的不同质量的图像;FIG2 is an image of different qualities provided by an embodiment of the present disclosure;
图3为本公开实施例提供的量子分类器的构建过程示意图;
FIG3 is a schematic diagram of a construction process of a quantum classifier provided in an embodiment of the present disclosure;
图4为本公开实施例提供的图像质量评价装置的结构组成示意图;FIG4 is a schematic diagram of the structure of an image quality assessment device provided by an embodiment of the present disclosure;
图5为本公开实施例提供的一种电子设备的结构组成示意图。FIG5 is a schematic diagram of the structural composition of an electronic device provided in an embodiment of the present disclosure.
实施方式Implementation
为使本公开实施例的目的、技术方案和优点更加清楚,下面将结合本公开实施例中附图,对本公开实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例是本公开一部分实施例,而不是全部的实施例。通常在此处附图中描述和示出的本公开实施例的组件可以以各种不同的配置来布置和设计。以下对在附图中提供的本公开的实施例的详细描述并非旨在限制要求保护的本公开的范围,而是表示本公开的选定实施例。基于本公开的实施例,本领域技术人员在没有做出创造性劳动的前提下所获得的所有其他实施例,都属于本公开保护的范围。In order to make the purpose, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, rather than all of the embodiments. The components of the embodiments of the present disclosure generally described and shown in the drawings here can be arranged and designed in various different configurations. The following detailed description of the embodiments of the present disclosure provided in the drawings is not intended to limit the scope of the present disclosure claimed for protection, but rather represents selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present disclosure.
应注意到:相似的标号和字母在下面的附图中表示类似项,一旦某一项在一个附图中被定义,则在随后的附图中不需要对其进行定义和解释。It should be noted that similar reference numerals and letters denote similar items in the following drawings, and once an item is defined in one drawing, it does not necessarily need to be defined or explained in the subsequent drawings.
本文中术语“和/或”,是描述一种关联关系,表示可以存在三种关系,例如,A和/或B,可以表示:单独存在A,同时存在A和B,单独存在B这三种情况。另外,本文中术语“至少一种”表示多种中的任意一种或多种中的至少两种的任意组合,例如,包括A、B、C中的至少一种,可以表示包括从A、B和C构成的集合中选择的任意一个或多个元素。The term "and/or" herein describes an association relationship, indicating that three relationships may exist. For example, A and/or B may indicate the existence of A alone, A and B at the same time, and B alone. In addition, the term "at least one" herein indicates any combination of at least two of any one or more of a plurality of. For example, at least one of A, B, and C may indicate any one or more elements selected from the set consisting of A, B, and C.
图像质量评价方法可以分为主观评价方法与客观评价方法。以下对主观评价方法和客观评价方法介绍进行介绍。Image quality evaluation methods can be divided into subjective evaluation methods and objective evaluation methods. The following introduces the subjective evaluation method and the objective evaluation method.
主观评价方法通过预先准备图像样本集,然后通过观测者对图像的观察对图像质量进行评价,得到主观评价结果,之后再通过对主观评价结果进行加工,通过设定一组图像中的主观评价值、均值和标准差综合作为对图像质量的评判依据。The subjective evaluation method prepares a set of image samples in advance, and then evaluates the image quality by observing the images to obtain the subjective evaluation results. The subjective evaluation results are then processed and the subjective evaluation value, mean and standard deviation in a group of images are set as the basis for judging the image quality.
客观评价方法分为全参考方法与半参考方法,其中,半参考方法结合机器学习的方法进行图像质量的评价,假设质量相近的图像在底层特征上具有相同的规律,不去对图像失真原因及设计特征的方法进行分析,单纯通过计算机学习得到的特征作为图像的评价标准。Objective evaluation methods are divided into full-reference methods and semi-reference methods. The semi-reference method combines machine learning methods to evaluate image quality. It assumes that images of similar quality have the same rules in underlying features. It does not analyze the causes of image distortion and the methods of designing features. It simply uses the features obtained by computer learning as the evaluation criteria for the image.
基于机器学习的图像质量评价方法面对的问题是如何在对图像质量进行评价时不要受到图像内容的干扰,尤其在对图像中提取特定特征统计量时,图像内容会对特征统计量带来较大的偏差;又或者在对内容很复杂的图像进行特征提取时,会比相同质量的内容简单的图像带来更多特征统计量,并且,由于没有实际的图像作为参考,为对多类别的图像进行评价平衡,容易使用相同的方法对复杂图像进行“过估计”,对简单的图像“欠估计”,导致对全体样本进行一致评价时存在问题。The problem faced by image quality assessment methods based on machine learning is how to evaluate image quality without being disturbed by image content, especially when extracting specific feature statistics from images, the image content will bring a large deviation to the feature statistics; or when extracting features from images with very complex content, it will bring more feature statistics than images of the same quality with simple content. In addition, since there are no actual images as references, in order to balance the evaluation of multiple categories of images, it is easy to use the same method to "over-estimate" complex images and "under-estimate" simple images, resulting in problems in making consistent evaluations of all samples.
本公开实施例的技术方案使用量子神经网络方法对图像进行分类,通过设计的量子神经网络可以对主观评价得到的图像进行较优图像的选择,相比较单纯使用主观评价值确定图像质量的方法,本公开实施例的技术方案有利于更快更优的进行较优图像的选取。The technical solution of the embodiment of the present disclosure uses a quantum neural network method to classify images. The designed quantum neural network can select better images from images obtained through subjective evaluation. Compared with the method of simply using subjective evaluation values to determine image quality, the technical solution of the embodiment of the present disclosure is conducive to faster and better selection of better images.
本公开实施例的技术方案可以利用量子编码方式将传统经典图像数据转换成量子数据编码,然后通过训练量子神经网络得出分类系统参数,通过筛选较优的分类系统参数得到图像质量筛选效果较优的量子分类器。之后,将图像的主观评价值作为量子分类器的输入数据,进而开展图像质量的分类工作,依据设定的阈值可自动筛选主观质量较优的待选图像。下面,对本公开实施例的图像质量评价方法介绍如下。The technical solution of the disclosed embodiment can use quantum coding to convert traditional classical image data into quantum data coding, and then obtain classification system parameters by training quantum neural network, and obtain quantum classifier with better image quality screening effect by screening better classification system parameters. After that, the subjective evaluation value of the image is used as the input data of the quantum classifier, and then the image quality classification work is carried out. According to the set threshold, the candidate images with better subjective quality can be automatically screened. The image quality evaluation method of the disclosed embodiment is introduced as follows.
请参阅图1,图1为本公开实施例提供的图像质量评价方法的流程示意图,如图1所示,本公开实施例的图像质量评价方法包括如下步骤:Please refer to FIG. 1, which is a flow chart of an image quality evaluation method provided by an embodiment of the present disclosure. As shown in FIG. 1, the image quality evaluation method of the embodiment of the present disclosure includes the following steps:
步骤101:构建量子分类器。Step 101: Build a quantum classifier.
本公开实施例中,量子分类器为对初始量子神经网络进行训练得到的一个对图像质量进行评价的量子分类器。In the disclosed embodiment, the quantum classifier is a quantum classifier for evaluating image quality obtained by training an initial quantum neural network.
在一种可能的实现方式中,上述步骤101包括如下步骤:In a possible implementation, step 101 includes the following steps:
1.1)获取图像数据集,将所述图像数据集中包括的各图像数据编码为对应的第一量子数据;1.1) Acquire an image data set, and encode each image data included in the image data set into corresponding first quantum data;
1.2)利用所述各图像数据对应的第一量子数据训练初始量子神经网络,得到量子分类器。1.2) Using the first quantum data corresponding to each image data to train an initial quantum neural network, a quantum classifier is obtained.
本公开实施例中,由于量子神经网络需要处理的数据是量子数据,对初始量子神经网络进行训练得到量子分类器时,需要将传统的数字图像数据转换为量子形式的数据,才能进行后续对初始量子神经网络的训练,得到用于对图像质量进行评价的量子分类器。In the disclosed embodiment, since the data that the quantum neural network needs to process is quantum data, when the initial quantum neural network is trained to obtain a quantum classifier, the traditional digital image data needs to be converted into data in quantum form before the subsequent training of the initial quantum neural network can be performed to obtain a quantum classifier for evaluating image quality.
在一种可能的实现方式中,所述图像数据集中的各图像数据为N维数据,N为大于等于1的整数,上述步骤1.1)包括如下步骤:In a possible implementation, each image data in the image data set is N-dimensional data, N is an integer greater than or equal to 1, and the above step 1.1) includes the following steps:
1.1.1)将所述图像数据集中的各图像数据设置为预设形式的图像数据;基于所述各图像数据对应的预设形式的图像数据确定初始状态的量子态数据;
1.1.1) setting each image data in the image data set as image data in a preset form; determining quantum state data of an initial state based on the image data in the preset form corresponding to each image data;
1.1.2)基于所述各图像数据对应的预设形式的图像数据,将所述各图像数据编码为量子门形式并作用在所述初始状态的量子态数据上,得到所述各图像数据对应的第一量子数据。1.1.2) Based on the image data in a preset format corresponding to each image data, each image data is encoded into a quantum gate format and acts on the quantum state data of the initial state to obtain the first quantum data corresponding to each image data.
以二维图像数据为例,在对初始量子神经网络进行训练之前,需要首先对二维图像数据进行编码,得到与各二维图像数据对应的量子数据。在一种实施方式中,编码过程如下:Taking two-dimensional image data as an example, before training the initial quantum neural network, the two-dimensional image data needs to be encoded first to obtain quantum data corresponding to each two-dimensional image data. In one embodiment, the encoding process is as follows:
设定二维数据的格式为则初始状态的量子态可设置为|xx>,二维图像数据{xk}可编码成量子门U(xk)的形式且作用在初始的量子态上,得到一系列如以下公式(1)的形式所示的量子数据,此表达过程即为将经典图像数据信息转变为量子数字信息的编码过程。
|ψ>k=U(xk)|xx> (1)Set the format of the two-dimensional data to Then the quantum state of the initial state can be set to |xx>, and the two-dimensional image data {x k } can be encoded into the form of a quantum gate U(x k ) and act on the initial quantum state to obtain a series of quantum data as shown in the following formula (1). This expression process is the encoding process of converting classical image data information into quantum digital information.
|ψ> k =U(x k )|xx> (1)
|ψ>k=U(xk)|xx> (1)Set the format of the two-dimensional data to Then the quantum state of the initial state can be set to |xx>, and the two-dimensional image data {x k } can be encoded into the form of a quantum gate U(x k ) and act on the initial quantum state to obtain a series of quantum data as shown in the following formula (1). This expression process is the encoding process of converting classical image data information into quantum digital information.
|ψ> k =U(x k )|xx> (1)
示例性的,若设置量子比特的数目为n,用n个量子比特编码二维的经典数字图像数据时,可以将量子门设定为以下公式(2)的形式:
For example, if the number of quantum bits is set to n, when two-dimensional classical digital image data is encoded with n quantum bits, the quantum gate can be set to the form of the following formula (2):
For example, if the number of quantum bits is set to n, when two-dimensional classical digital image data is encoded with n quantum bits, the quantum gate can be set to the form of the following formula (2):
以经典数字数据x=(x0,x1)=(1,0)为例,用2比特的构造的量子门可表达为以下公式(3)的形式:
Taking the classical digital data x = (x 0 , x 1 ) = (1, 0) as an example, the quantum gate constructed with 2 bits can be expressed in the form of the following formula (3):
Taking the classical digital data x = (x 0 , x 1 ) = (1, 0) as an example, the quantum gate constructed with 2 bits can be expressed in the form of the following formula (3):
将x=(x0,x1)=(1,0)代入上述公式(3)中,得到如下公式(4)所示的量子门U(x):
Substituting x=(x 0 ,x 1 )=(1,0) into the above formula (3), we obtain the quantum gate U(x) shown in the following formula (4):
Substituting x=(x 0 ,x 1 )=(1,0) into the above formula (3), we obtain the quantum gate U(x) shown in the following formula (4):
在一种可能的实现方式中,上述步骤1.2)包括如下步骤:In a possible implementation, the above step 1.2) includes the following steps:
1.2.1)将所述各图像数据对应的第一量子数据输入至初始量子神经网络,得到第二量子数据;1.2.1) Inputting the first quantum data corresponding to each image data into the initial quantum neural network to obtain second quantum data;
1.2.2)利用预设测量方法测量所述第二量子数据,得到所述第二量子数据对应的第一测量数据;1.2.2) measuring the second quantum data using a preset measurement method to obtain first measurement data corresponding to the second quantum data;
1.2.3)对所述第一测量数据进行偏置处理得到位于预设映射区间的第二测量数据;1.2.3) performing offset processing on the first measurement data to obtain second measurement data located in a preset mapping interval;
1.2.4)基于所述第二测量数据以及所述第一量子数据确定所述初始量子神经网络的损失函数;1.2.4) determining a loss function of the initial quantum neural network based on the second measurement data and the first quantum data;
1.2.5)确定所述损失函数是否满足第二条件;在所述损失函数不满足所述第二条件的情况下,重复执行步骤1.2.1)至步骤1.2.4),直至所述损失函数满足所述第二条件。1.2.5) Determine whether the loss function satisfies the second condition; if the loss function does not satisfy the second condition, repeat steps 1.2.1) to 1.2.4) until the loss function satisfies the second condition.
本公开实施例中,通过步骤1.1)得到的第一量子数据为又可以称之为量子输入态数据,即为待向初始量子神经网络中输入的量子数据,将第一量子数据输入初始量子神经网络中后,得到量子输出态数据。In the disclosed embodiment, the first quantum data obtained through step 1.1) can also be called quantum input state data, that is, the quantum data to be input into the initial quantum neural network. After the first quantum data is input into the initial quantum neural network, quantum output state data is obtained.
将整个初始量子神经网络设置为U(θ),以如下公式(5)的矩阵形式作为中间旋转门:
The entire initial quantum neural network is set to U(θ), and the matrix form of the following formula (5) is used as the intermediate revolving gate:
The entire initial quantum neural network is set to U(θ), and the matrix form of the following formula (5) is used as the intermediate revolving gate:
上述公式(4)得到的2个比特的量子门矩阵可以表达为如下公式(6)的形式:
The 2-bit quantum gate matrix obtained by the above formula (4) can be expressed in the form of the following formula (6):
The 2-bit quantum gate matrix obtained by the above formula (4) can be expressed in the form of the following formula (6):
将量子态输入初始量子神经网络,通过初始量子神经网络计算后的量子输出态为|ψ>,|ψ>=U(θ)|ψ>,量子输出态|ψ>即对应步骤1.2.1)得到的第二量子数据。The quantum state is input into the initial quantum neural network. The quantum output state calculated by the initial quantum neural network is |ψ>, |ψ>=U(θ)|ψ>, and the quantum output state |ψ> corresponds to the second quantum data obtained in step 1.2.1).
针对上述步骤1.2.2),由于通过步骤1.2.1)中初始量子神经网络处理过的量子态会发生变化,需要重新测量输出的量子态。在一种实施方式中,可以采用测量泡利Z算符的方式将上述定义的量子比特序列中的第一个作为期望值,可定义为
Regarding the above step 1.2.2), since the quantum state processed by the initial quantum neural network in step 1.2.1) will change, it is necessary to re-measure the output quantum state. In one embodiment, the first of the above-defined quantum bit sequences can be used as the expected value by measuring the Pauli Z operator, which can be defined as
需要说明的是,本公开实施例中的预设测量方法包括但不限于泡利Z算符测量方式,还可以是任何能测量出第二量子数据的测量方式,本公开实施例不作限定。It should be noted that the preset measurement method in the embodiments of the present disclosure includes but is not limited to the Pauli Z operator measurement method, and can also be any measurement method that can measure the second quantum data, which is not limited in the embodiments of the present disclosure.
针对上述步骤1.2.3)及步骤1.2.4),示例性的,可以规定测量后的<Z>的取值范围为[-1,1],相应的可以定义为平方损失函数作为对初始量子神经网络进行训练时损失函数的映射区间,使得量子神经网络训练过程中损失函数的范围在[0,1]之间。For the above steps 1.2.3) and 1.2.4), for example, the value range of <Z> after measurement can be specified as [-1, 1], and correspondingly, The square loss function is used as the mapping interval of the loss function when training the initial quantum neural network, so that the range of the loss function during the quantum neural network training process is between [0,1].
对于上述步骤1.2.5),在一种实施方式中,确定损失函数是否满足第二条件,可以为确定损失函数是否小于设置的数值,如0.001,在损失函数小于设置的数值的情况下,即认为量子神经网络已训练完成。在另一种实施方式中,确定损失函数是否满足第二条件,可以为确定损失函数是否为最小化损失函数,在损失函数为最小化损失函数的情况下,即认为量子神经网络已训练完成。For the above step 1.2.5), in one embodiment, determining whether the loss function satisfies the second condition may be determining whether the loss function is less than a set value, such as 0.001. In the case where the loss function is less than the set value, the quantum neural network is considered to have been trained. In another embodiment, determining whether the loss function satisfies the second condition may be determining whether the loss function is a minimized loss function. In the case where the loss function is a minimized loss function, the quantum neural network is considered to have been trained.
在损失函数不满足第二条件的情况下,则需要更新量子神经网络的训练参数,并利用第一量子数据重复训练量子神经网络,直到训练得到的神经网络的损失函数满足第二条件。When the loss function does not satisfy the second condition, it is necessary to update the training parameters of the quantum neural network and repeatedly train the quantum neural network using the first quantum data until the loss function of the trained neural network satisfies the second condition.
在一种可能的实现方式中,上述步骤1.2)还包括如下步骤:In a possible implementation, the above step 1.2) further includes the following steps:
1.2.6)在所述损失函数满足第二条件的情况下,将所述损失函数对应的所述初始量子神经网络的参数确定为目标参数,并基于所述目标参数得到所述量子分类器。1.2.6) When the loss function satisfies the second condition, the parameters of the initial quantum neural network corresponding to the loss function are determined as target parameters, and the quantum classifier is obtained based on the target parameters.
在损失函数满足第二条件的情况下,则认为量子神经网络已训练完成,此时可得到训练完成的量子神经网络对应的网络参数,取该网络参数得到的量子神经网络,即为用于对图像质量进行评价的量子分类器。利用得到的量子分类器即可对一组待进行图像质量评价的图像进行图像质量的评价。When the loss function satisfies the second condition, the quantum neural network is considered to have been trained. At this time, the network parameters corresponding to the trained quantum neural network can be obtained. The quantum neural network obtained by taking the network parameters is the quantum classifier used to evaluate the image quality. The obtained quantum classifier can be used to evaluate the image quality of a group of images to be evaluated.
步骤102:将多个待评价图像对应的主观评价值输入所述量子分类器,利用所述量子分类器确定所述多个待评价图像中图像质量满足第一条件的目标图像。Step 102: Inputting subjective evaluation values corresponding to a plurality of images to be evaluated into the quantum classifier, and using the quantum classifier to determine a target image whose image quality meets a first condition among the plurality of images to be evaluated.
本公开实施例中,针对所述多个待评价图像中的每个待评价图像,该待评价图像对应的主观评价值基于至少一个图像观测者对该待评价图像的图像质量评分得到。In the embodiment of the present disclosure, for each image to be evaluated among the multiple images to be evaluated, the subjective evaluation value corresponding to the image to be evaluated is obtained based on the image quality score of the image to be evaluated by at least one image observer.
在一种实施方式中,可以通过图像观测者的主观打分来判断图像质量。可以采用的方式为:让图像观测者根据事先规定好的评价准则或者自己的主观经验,对待评价图像按照视觉效果做出质量判断,并给出质量分数,即给出图像质量评分,对所有图像观测者给出的质量分数进行加权平均,得到的结果就是该幅图像的主观评价值,也可以称之为主观评分值,主观评价值越高则说明图像的质量越好。In one implementation, the image quality can be judged by subjective scoring by image observers. The method that can be adopted is: let the image observer make a quality judgment on the image to be evaluated according to the visual effect based on the evaluation criteria specified in advance or his own subjective experience, and give a quality score, that is, give an image quality score, and perform a weighted average of the quality scores given by all image observers. The result obtained is the subjective evaluation value of the image, which can also be called a subjective scoring value. The higher the subjective evaluation value, the better the image quality.
本公开实施例中,多个待评价图像中图像质量满足第一条件的图像,可以为多个待评价图像中图像质量较优的图像,即图像质量排序靠前的图像。In the embodiment of the present disclosure, the image whose image quality satisfies the first condition among the multiple images to be evaluated may be an image with better image quality among the multiple images to be evaluated, that is, an image with a higher image quality ranking.
本公开实施例中,利用得到的量子分类器即可对一组待进行图像质量评价的图像进行图像质量的评价,例如,可以将一组待进行图像质量评价的多幅图像对应的主观评价值输入至训练得到的量子分类器中,利用量子分类器筛选出一组图像中图像质量较优的图像。In the disclosed embodiment, the obtained quantum classifier can be used to evaluate the image quality of a group of images to be evaluated. For example, the subjective evaluation values corresponding to a group of multiple images to be evaluated can be input into the trained quantum classifier, and the quantum classifier can be used to screen out images with better image quality from the group of images.
在一种可能的实现方式中,上述步骤102可通过如下步骤实现:In a possible implementation, the above step 102 may be implemented by the following steps:
利用所述量子分类器产生的决策边界确定所述多个待评价图像中的各待评价图像在所述决策边界中所处的区域,将位于所述决策边界的中心区域的主观评价值对应的待评价图像确定为所述多个待评价图像中图像质量满足第一条件的目标图像。The decision boundary generated by the quantum classifier is used to determine the area where each image to be evaluated among the multiple images to be evaluated is located in the decision boundary, and the image to be evaluated corresponding to the subjective evaluation value located in the central area of the decision boundary is determined as the target image whose image quality meets the first condition among the multiple images to be evaluated.
在一种实施方式中,可以选取图像质量主观评价值数据集,利用步骤101构建的量子分类器,将图像质量主观评价值数据集输入至已构建好的量子分类器中,依据量子分类器产生的决策边界来
区分不同图像所在区域,在量子决策边界中心区域dmos值(即主观评价值)对应图像即为质量较优图像。下面表1为一组图像质量评分数据集。In one implementation, a subjective image quality evaluation value dataset can be selected, and the image quality subjective evaluation value dataset can be input into the constructed quantum classifier using the quantum classifier constructed in step 101, and the decision boundary generated by the quantum classifier can be used to determine the subjective image quality evaluation value dataset. Different image regions are distinguished, and the image corresponding to the dmos value (i.e., subjective evaluation value) in the center of the quantum decision boundary is the image with better quality. Table 1 below is a set of image quality scoring data sets.
表1图像质量评分数据集
Table 1 Image quality rating dataset
Table 1 Image quality rating dataset
表1中image代表图像编号,dst_idx代表图像编码方式,dst_lev图像等级,dmos_std代表平均主观得分差,dmos代表主观评价值。In Table 1, image represents the image number, dst_idx represents the image encoding method, dst_lev represents the image level, dmos_std represents the mean subjective score difference, and dmos represents the subjective evaluation value.
图2为本公开实施例提供的不同质量的图像,图中的(a)~(e)共5幅图像分别具有不同的主
观评价值,将图2中5幅图像的主观评价值输入训练好的量子分类器,依据各图像不同的主观评价值可筛选出图2中最左侧的图像(a)为图像视觉质量最好的图像。图2中,5幅图像的图像质量从左到右依次降低。FIG2 shows images of different qualities provided by an embodiment of the present disclosure. The five images (a) to (e) in the figure have different main The subjective evaluation values of the five images in Figure 2 are input into the trained quantum classifier. According to the different subjective evaluation values of each image, the leftmost image (a) in Figure 2 can be selected as the image with the best visual quality. In Figure 2, the image quality of the five images decreases from left to right.
本公开能够利用量子神经网络生成的量子分类器通过对图像主观评价值进行分类,从多幅图像中筛选出质量较优的图像。The present invention can utilize a quantum classifier generated by a quantum neural network to classify subjective evaluation values of images and select images with better quality from multiple images.
本公开实施例提供的图像质量的评价方法不同于传统机器学习方法进行分类器的生成,本公开实施例使用一种量子神经网络方法生成分类器,与传统的图像筛选方法实现的基本原理是不同的。本公开实施例的技术方案利用量子神经网络生成的分类器通过对图像主观评价值进行分类,筛选出质量较优的图像,具有较好的应用价值。另外,本公开实施例的技术方案不同于传统方法在经典计算机中运行,本公开实施例的方案可适配在电子计算机与量子计算机模式形态中运行,对于未来使用量子模态的计算机,具有较高的前沿技术预研价值。The image quality evaluation method provided by the embodiment of the present disclosure is different from the traditional machine learning method for generating a classifier. The embodiment of the present disclosure uses a quantum neural network method to generate a classifier, which is different from the basic principle of the traditional image screening method. The technical solution of the embodiment of the present disclosure uses a classifier generated by a quantum neural network to classify the subjective evaluation values of the image and screen out images with better quality, which has good application value. In addition, the technical solution of the embodiment of the present disclosure is different from the traditional method running in a classical computer. The solution of the embodiment of the present disclosure can be adapted to run in electronic computers and quantum computer modes, and has a high value of frontier technology pre-research for computers using quantum modes in the future.
请参阅图3,图3为本公开实施例提供的量子分类器的构建过程示意图,图3涉及图像数据的处理过程以及量子神经网络的训练过程。如图3所示,量子分类器的生成过程分为量子数据编码和量子神经网络训练两个部分。其中,量子数据编码部分将经典的数字图像信息转换为量子数据信息;量子神经网络训练得到分类器的过程,是基于得到的量子数据信息对初始的量子神经网络进行训练,得到量子分类器的分类系统参数,从而得到量子分类器。量子数据编码过程以及量子神经网络训练得到量子分类器的过程可参照上述步骤101。Please refer to Figure 3, which is a schematic diagram of the construction process of the quantum classifier provided by the embodiment of the present disclosure. Figure 3 involves the processing process of image data and the training process of the quantum neural network. As shown in Figure 3, the generation process of the quantum classifier is divided into two parts: quantum data encoding and quantum neural network training. Among them, the quantum data encoding part converts the classical digital image information into quantum data information; the process of obtaining the classifier by quantum neural network training is to train the initial quantum neural network based on the obtained quantum data information to obtain the classification system parameters of the quantum classifier, thereby obtaining the quantum classifier. The process of quantum data encoding and the process of obtaining the quantum classifier by quantum neural network training can refer to the above step 101.
请参阅图4,图4为本公开实施例提供的图像质量评价装置的结构组成示意图,如图4所示,所述装置包括:Please refer to FIG. 4 , which is a schematic diagram of the structure of an image quality assessment device provided by an embodiment of the present disclosure. As shown in FIG. 4 , the device includes:
构建单元401,配置为构建量子分类器;A construction unit 401 is configured to construct a quantum classifier;
确定单元402,配置为将多个待评价图像对应的主观评价值输入所述量子分类器,利用所述量子分类器确定所述多个待评价图像中图像质量满足第一条件的目标图像;A determination unit 402 is configured to input subjective evaluation values corresponding to a plurality of images to be evaluated into the quantum classifier, and use the quantum classifier to determine a target image whose image quality meets a first condition among the plurality of images to be evaluated;
其中,针对所述多个待评价图像中的每个待评价图像,该待评价图像对应的主观评价值基于至少一个图像观测者对该待评价图像的图像质量评分得到。For each image to be evaluated among the multiple images to be evaluated, the subjective evaluation value corresponding to the image to be evaluated is obtained based on an image quality score of the image to be evaluated by at least one image observer.
在一种可能的实现方式中,所述构建单元401,配置为获取图像数据集,将所述图像数据集中包括的各图像数据编码为对应的第一量子数据;利用所述各图像数据对应的第一量子数据训练初始量子神经网络,得到量子分类器。In a possible implementation, the construction unit 401 is configured to obtain an image data set, encode each image data included in the image data set into corresponding first quantum data; and train an initial quantum neural network using the first quantum data corresponding to each image data to obtain a quantum classifier.
在一种可能的实现方式中,所述构建单元401,配置为将所述各图像数据对应的第一量子数据输入至初始量子神经网络,得到第二量子数据;利用预设测量方法测量所述第二量子数据,得到所述第二量子数据对应的第一测量数据;对所述第一测量数据进行偏置处理得到位于预设映射区间的第二测量数据;基于所述第二测量数据以及所述第一量子数据确定所述初始量子神经网络的损失函数;确定所述损失函数是否满足第二条件;在所述损失函数不满足所述第二条件的情况下,重复执行如下步骤,直至所述损失函数满足所述第二条件:将所述各图像数据对应的第一量子数据输入至初始量子神经网络,得到第二量子数据;利用预设测量方法测量所述第二量子数据,得到所述第二量子数据对应的第一测量数据;对所述第一测量数据进行偏置处理得到位于预设映射区间的第二测量数据;基于所述第二测量数据以及所述第一量子数据确定所述初始量子神经网络的损失函数。In a possible implementation, the construction unit 401 is configured to input the first quantum data corresponding to each image data into the initial quantum neural network to obtain second quantum data; measure the second quantum data using a preset measurement method to obtain first measurement data corresponding to the second quantum data; bias the first measurement data to obtain second measurement data located in a preset mapping interval; determine the loss function of the initial quantum neural network based on the second measurement data and the first quantum data; determine whether the loss function satisfies a second condition; if the loss function does not satisfy the second condition, repeat the following steps until the loss function satisfies the second condition: input the first quantum data corresponding to each image data into the initial quantum neural network to obtain second quantum data; measure the second quantum data using a preset measurement method to obtain first measurement data corresponding to the second quantum data; bias the first measurement data to obtain second measurement data located in a preset mapping interval; determine the loss function of the initial quantum neural network based on the second measurement data and the first quantum data.
在一种可能的实现方式中,所述构建单元401,配置为在所述损失函数满足第二条件的情况下,将所述损失函数对应的所述初始量子神经网络的参数确定为目标参数,并基于所述目标参数得到所述量子分类器。In a possible implementation, the construction unit 401 is configured to, when the loss function satisfies the second condition, determine the parameters of the initial quantum neural network corresponding to the loss function as target parameters, and obtain the quantum classifier based on the target parameters.
在一种可能的实现方式中,所述图像数据集中的各图像数据为N维数据,N为大于等于1的整数;所述构建单元401,配置为将所述图像数据集中的各图像数据设置为预设形式的图像数据;基于所述各图像数据对应的预设形式的图像数据确定初始状态的量子态数据;基于所述各图像数据对应的预设形式的图像数据,将所述各图像数据编码为量子门形式并作用在所述初始状态的量子态数据上,得到所述各图像数据对应的第一量子数据。In a possible implementation, each image data in the image data set is N-dimensional data, where N is an integer greater than or equal to 1; the construction unit 401 is configured to set each image data in the image data set as image data in a preset form; determine quantum state data of an initial state based on the image data in the preset form corresponding to each image data; based on the image data in the preset form corresponding to each image data, encode each image data into a quantum gate form and act on the quantum state data of the initial state to obtain first quantum data corresponding to each image data.
在一种可能的实现方式中,所述确定单元402,配置为利用所述量子分类器产生的决策边界确定所述多个待评价图像中的各待评价图像在所述决策边界中所处的区域,将位于所述决策边界的中心区域的主观评价值对应的待评价图像确定为所述多个待评价图像中图像质量满足第一条件的目标图像。In a possible implementation, the determination unit 402 is configured to use the decision boundary generated by the quantum classifier to determine the area where each image to be evaluated among the multiple images to be evaluated is located in the decision boundary, and determine the image to be evaluated corresponding to the subjective evaluation value located in the central area of the decision boundary as the target image whose image quality meets the first condition among the multiple images to be evaluated.
本领域技术人员应当理解,图4所示的图像质量评价装置中的各单元的实现功能可参照前述图像质量评价方法的相关描述而理解。图4所示的图像质量评价装置中的各单元的功能可通过运行于处理器上的程序而实现,也可通过逻辑电路而实现。
Those skilled in the art should understand that the functions of each unit in the image quality assessment device shown in FIG4 can be understood by referring to the relevant description of the above-mentioned image quality assessment method. The functions of each unit in the image quality assessment device shown in FIG4 can be implemented by a program running on a processor, or by a logic circuit.
本公开实施例还提供了一种电子设备。图5为本公开实施例的电子设备的硬件结构示意图,如图5所示,电子设备包括:配置为进行数据传输的通信组件503、至少一个处理器501和配置为存储能够在处理器501上运行的计算机程序的存储器502。终端中的各个组件通过总线系统504耦合在一起。可理解,总线系统504配置为实现这些组件之间的连接通信。总线系统504除包括数据总线之外,还包括电源总线、控制总线和状态信号总线。但是为了清楚说明起见,在图5中将各种总线都标为总线系统504。The embodiment of the present disclosure also provides an electronic device. FIG5 is a schematic diagram of the hardware structure of the electronic device of the embodiment of the present disclosure. As shown in FIG5 , the electronic device includes: a communication component 503 configured to perform data transmission, at least one processor 501, and a memory 502 configured to store a computer program that can be run on the processor 501. The various components in the terminal are coupled together through a bus system 504. It can be understood that the bus system 504 is configured to achieve connection and communication between these components. In addition to the data bus, the bus system 504 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, various buses are marked as bus systems 504 in FIG5.
其中,所述处理器501执行所述计算机程序时至少执行图1所示的方法的步骤。When the processor 501 executes the computer program, it at least performs the steps of the method shown in FIG. 1 .
可以理解,存储器502可以是易失性存储器或非易失性存储器,也可包括易失性和非易失性存储器两者。其中,非易失性存储器可以是只读存储器(ROM,Read Only Memory)、可编程只读存储器(PROM,Programmable Read-Only Memory)、可擦除可编程只读存储器(EPROM,Erasable Programmable Read-Only Memory)、电可擦除可编程只读存储器(EEPROM,Electrically Erasable Programmable Read-Only Memory)、磁性随机存取存储器(FRAM,ferromagnetic random access memory)、快闪存储器(Flash Memory)、磁表面存储器、光盘、或只读光盘(CD-ROM,Compact Disc Read-Only Memory);磁表面存储器可以是磁盘存储器或磁带存储器。易失性存储器可以是随机存取存储器(RAM,Random Access Memory),其用作外部高速缓存。通过示例性但不是限制性说明,许多形式的RAM可用,例如静态随机存取存储器(SRAM,Static Random Access Memory)、同步静态随机存取存储器(SSRAM,Synchronous Static Random Access Memory)、动态随机存取存储器(DRAM,Dynamic Random Access Memory)、同步动态随机存取存储器(SDRAM,Synchronous Dynamic Random Access Memory)、双倍数据速率同步动态随机存取存储器(DDRSDRAM,Double Data Rate Synchronous Dynamic Random Access Memory)、增强型同步动态随机存取存储器(ESDRAM,Enhanced Synchronous Dynamic Random Access Memory)、同步连接动态随机存取存储器(SLDRAM,SyncLink Dynamic Random Access Memory)、直接内存总线随机存取存储器(DRRAM,Direct Rambus Random Access Memory)。本公开实施例描述的存储器502旨在包括但不限于这些和任意其它适合类型的存储器。It can be understood that the memory 502 can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM); the magnetic surface memory can be a disk memory or a tape memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), direct memory bus random access memory (DRRAM). The memory 502 described in the embodiments of the present disclosure is intended to include, but is not limited to, these and any other suitable types of memory.
上述本公开实施例揭示的方法可以应用于处理器501中,或者由处理器501实现。处理器501可能是一种集成电路芯片,具有信号的处理能力。在实现过程中,上述方法的各步骤可以通过处理器501中的硬件的集成逻辑电路或者软件形式的指令完成。上述的处理器501可以是通用处理器、DSP,或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。处理器501可以实现或者执行本公开实施例中的公开的各方法、步骤及逻辑框图。通用处理器可以是微处理器或者任何常规的处理器等。结合本公开实施例所公开的方法的步骤,可以直接体现为硬件译码处理器执行完成,或者用译码处理器中的硬件及软件模块组合执行完成。软件模块可以位于存储介质中,该存储介质位于存储器502,处理器501读取存储器502中的信息,结合其硬件完成前述方法的步骤。The method disclosed in the above embodiment of the present disclosure can be applied to the processor 501, or implemented by the processor 501. The processor 501 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit in the processor 501 or the instruction in the form of software. The above processor 501 can be a general-purpose processor, a DSP, or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The processor 501 can implement or execute the methods, steps and logic block diagrams disclosed in the embodiment of the present disclosure. The general-purpose processor can be a microprocessor or any conventional processor, etc. The steps of the method disclosed in the embodiment of the present disclosure can be directly embodied as a hardware decoding processor to execute, or the hardware and software modules in the decoding processor can be combined to execute. The software module can be located in a storage medium, which is located in the memory 502. The processor 501 reads the information in the memory 502 and completes the steps of the above method in combination with its hardware.
在示例性实施例中,电子设备可以被一个或多个应用专用集成电路(ASIC,Application Specific Integrated Circuit)、DSP、可编程逻辑器件(PLD,Programmable Logic Device)、复杂可编程逻辑器件(CPLD,Complex Programmable Logic Device)、FPGA、通用处理器、控制器、MCU、微处理器(Microprocessor)、或其他电子元件实现,用于执行前述的图像质量评价方法。In an exemplary embodiment, the electronic device can be implemented by one or more application specific integrated circuits (ASIC), DSP, programmable logic device (PLD), complex programmable logic device (CPLD), FPGA, general-purpose processor, controller, MCU, microprocessor, or other electronic components to execute the aforementioned image quality evaluation method.
本公开实施例还提供一种计算机可读存储介质,其上存储有计算机程序,该程序被处理器执行时至少用于执行图1所示方法的步骤。所述计算机可读存储介质可以为存储器。所述存储器可以为如图5所示的存储器502。The embodiment of the present disclosure further provides a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, it is used to at least perform the steps of the method shown in Figure 1. The computer-readable storage medium may be a memory. The memory may be the memory 502 shown in Figure 5.
本公开实施例提供了一种计算机程序产品,包括计算机可读代码,在所述计算机可读代码在电子设备中运行的情况下,所述电子设备中的处理器执行上述的图像质量评价方法。An embodiment of the present disclosure provides a computer program product, including a computer-readable code. When the computer-readable code is executed in an electronic device, a processor in the electronic device executes the above-mentioned image quality assessment method.
本公开实施例所记载的技术方案之间,在不冲突的情况下,可以任意组合。The technical solutions described in the embodiments of the present disclosure can be combined arbitrarily without conflict.
在本公开所提供的几个实施例中,应该理解到,所揭露的方法和智能设备,可以通过其它的方式实现。以上所描述的设备实施例是示意性的,例如,所述单元的划分,为一种逻辑功能划分,实际实现时可以有另外的划分方式,如:多个单元或组件可以结合,或可以集成到另一个系统,或一些特征可以忽略,或不执行。另外,所显示或讨论的各组成部分相互之间的耦合、或直接耦合、或通信连接可以是通过一些接口,设备或单元的间接耦合或通信连接,可以是电性的、机械的或其它形式的。In the several embodiments provided in the present disclosure, it should be understood that the disclosed methods and intelligent devices can be implemented in other ways. The device embodiments described above are schematic. For example, the division of the units is a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.
上述作为分离部件说明的单元可以是、或也可以不是物理上分开的,作为单元显示的部件可以是、或也可以不是物理单元,即可以位于一个地方,也可以分布到多个网络单元上;可以根据实际的需要选择其中的部分或全部单元来实现本实施例方案的目的。
The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.
另外,在本公开各实施例中的各功能单元可以全部集成在一个第二处理单元中,也可以是各单元分别单独作为一个单元,也可以两个或两个以上单元集成在一个单元中;上述集成的单元既可以采用硬件的形式实现,也可以采用硬件加软件功能单元的形式实现。In addition, all functional units in the embodiments of the present disclosure may be integrated into a second processing unit, or each unit may be separately configured as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.
以上所述,仅为本公开的实施方式,但本公开的保护范围并不局限于此,任何熟悉本技术领域的技术人员在本公开揭露的技术范围内,可轻易想到变化或替换,都应涵盖在本公开的保护范围之内。The above description is only an implementation mode of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present disclosure, which should be included in the protection scope of the present disclosure.
通过构建量子分类器;将多个待评价图像对应的主观评价值输入所述量子分类器,利用所述量子分类器确定所述多个待评价图像中图像质量满足第一条件的目标图像,能够利用量子神经网络生成的量子分类器通过对图像主观评价值进行分类,从多幅图像中筛选出质量较优的图像。
By constructing a quantum classifier; inputting the subjective evaluation values corresponding to a plurality of images to be evaluated into the quantum classifier, and using the quantum classifier to determine a target image whose image quality meets a first condition among the plurality of images to be evaluated, the quantum classifier generated by the quantum neural network can be used to classify the subjective evaluation values of the images, thereby screening out images with better quality from the plurality of images.
Claims (15)
- 一种图像质量评价方法,所述方法包括:A method for evaluating image quality, the method comprising:构建量子分类器;Building a quantum classifier;将多个待评价图像对应的主观评价值输入所述量子分类器,利用所述量子分类器确定所述多个待评价图像中图像质量满足第一条件的目标图像;Inputting subjective evaluation values corresponding to a plurality of images to be evaluated into the quantum classifier, and using the quantum classifier to determine a target image among the plurality of images to be evaluated whose image quality meets a first condition;其中,针对所述多个待评价图像中的每个待评价图像,该待评价图像对应的主观评价值基于至少一个图像观测者对该待评价图像的图像质量评分得到。For each image to be evaluated among the multiple images to be evaluated, the subjective evaluation value corresponding to the image to be evaluated is obtained based on an image quality score of the image to be evaluated by at least one image observer.
- 根据权利要求1所述的方法,其中,所述构建量子分类器,包括:The method according to claim 1, wherein said constructing a quantum classifier comprises:获取图像数据集,将所述图像数据集中包括的各图像数据编码为对应的第一量子数据;Acquire an image data set, and encode each image data included in the image data set into corresponding first quantum data;利用所述各图像数据对应的第一量子数据训练初始量子神经网络,得到量子分类器。The initial quantum neural network is trained using the first quantum data corresponding to each image data to obtain a quantum classifier.
- 根据权利要求2所述的方法,其中,所述利用所述各图像数据对应的第一量子数据训练初始量子神经网络,得到量子分类器,包括:The method according to claim 2, wherein the step of training an initial quantum neural network using the first quantum data corresponding to each image data to obtain a quantum classifier comprises:将所述各图像数据对应的第一量子数据输入至初始量子神经网络,得到第二量子数据;Inputting the first quantum data corresponding to each image data into the initial quantum neural network to obtain second quantum data;利用预设测量方法测量所述第二量子数据,得到所述第二量子数据对应的第一测量数据;Measuring the second quantum data using a preset measurement method to obtain first measurement data corresponding to the second quantum data;对所述第一测量数据进行偏置处理得到位于预设映射区间的第二测量数据;Performing offset processing on the first measurement data to obtain second measurement data located in a preset mapping interval;基于所述第二测量数据以及所述第一量子数据确定所述初始量子神经网络的损失函数;Determine a loss function of the initial quantum neural network based on the second measurement data and the first quantum data;确定所述损失函数是否满足第二条件;在所述损失函数不满足所述第二条件的情况下,重复执行如下步骤,直至所述损失函数满足所述第二条件:Determine whether the loss function satisfies a second condition; if the loss function does not satisfy the second condition, repeat the following steps until the loss function satisfies the second condition:将所述各图像数据对应的第一量子数据输入至初始量子神经网络,得到第二量子数据;Inputting the first quantum data corresponding to each image data into the initial quantum neural network to obtain second quantum data;利用预设测量方法测量所述第二量子数据,得到所述第二量子数据对应的第一测量数据;Measuring the second quantum data using a preset measurement method to obtain first measurement data corresponding to the second quantum data;对所述第一测量数据进行偏置处理得到位于预设映射区间的第二测量数据;Performing offset processing on the first measurement data to obtain second measurement data located in a preset mapping interval;基于所述第二测量数据以及所述第一量子数据确定所述初始量子神经网络的损失函数。A loss function of the initial quantum neural network is determined based on the second measurement data and the first quantum data.
- 根据权利要求3所述的方法,其中,所述利用所述各图像数据对应的第一量子数据训练初始量子神经网络,得到量子分类器,还包括:The method according to claim 3, wherein the step of training an initial quantum neural network using the first quantum data corresponding to each image data to obtain a quantum classifier further comprises:在所述损失函数满足第二条件的情况下,将所述损失函数对应的所述初始量子神经网络的参数确定为目标参数,并基于所述目标参数得到所述量子分类器。When the loss function satisfies the second condition, the parameters of the initial quantum neural network corresponding to the loss function are determined as target parameters, and the quantum classifier is obtained based on the target parameters.
- 根据权利要求2至4中任一项所述的方法,其中,所述图像数据集中的各图像数据为N维数据,N为大于等于1的整数;所述将所述图像数据集中包括的各图像数据编码为对应的第一量子数据,包括:The method according to any one of claims 2 to 4, wherein each image data in the image data set is N-dimensional data, and N is an integer greater than or equal to 1; encoding each image data included in the image data set into corresponding first quantum data comprises:将所述图像数据集中的各图像数据设置为预设形式的图像数据;基于所述各图像数据对应的预设形式的图像数据确定初始状态的量子态数据;Setting each image data in the image data set to image data in a preset form; determining quantum state data of an initial state based on the image data in the preset form corresponding to each image data;基于所述各图像数据对应的预设形式的图像数据,将所述各图像数据编码为量子门形式并作用在所述初始状态的量子态数据上,得到所述各图像数据对应的第一量子数据。Based on the image data in a preset form corresponding to each image data, each image data is encoded into a quantum gate form and acts on the quantum state data of the initial state to obtain the first quantum data corresponding to each image data.
- 根据权利要求1至4中任一项所述的方法,其中,所述利用所述量子分类器确定所述多个待评价图像中图像质量满足第一条件的目标图像,包括:The method according to any one of claims 1 to 4, wherein the step of using the quantum classifier to determine a target image whose image quality satisfies a first condition among the multiple images to be evaluated comprises:利用所述量子分类器产生的决策边界确定所述多个待评价图像中的各待评价图像在所述决策边界中所处的区域,将位于所述决策边界的中心区域的主观评价值对应的待评价图像确定为所述多个待评价图像中图像质量满足第一条件的目标图像。The decision boundary generated by the quantum classifier is used to determine the area where each image to be evaluated among the multiple images to be evaluated is located in the decision boundary, and the image to be evaluated corresponding to the subjective evaluation value located in the central area of the decision boundary is determined as the target image whose image quality meets the first condition among the multiple images to be evaluated.
- 一种图像质量评价装置,所述装置包括:An image quality evaluation device, comprising:构建单元,配置为构建量子分类器;a building unit configured to build a quantum classifier;确定单元,配置为将多个待评价图像对应的主观评价值输入所述量子分类器,利用所述量子分类器确定所述多个待评价图像中图像质量满足第一条件的目标图像;A determination unit is configured to input subjective evaluation values corresponding to a plurality of images to be evaluated into the quantum classifier, and use the quantum classifier to determine a target image whose image quality meets a first condition among the plurality of images to be evaluated;其中,针对所述多个待评价图像中的每个待评价图像,该待评价图像对应的主观评价值基于至少一个图像观测者对该待评价图像的图像质量评分得到。For each image to be evaluated among the multiple images to be evaluated, the subjective evaluation value corresponding to the image to be evaluated is obtained based on an image quality score of the image to be evaluated by at least one image observer.
- 根据权利要求7所述的装置,其中,构建单元,配置为获取图像数据集,将所述图像数据集中包括的各图像数据编码为对应的第一量子数据;利用所述各图像数据对应的第一量子数据训练初始量子神经网络,得到量子分类器。The device according to claim 7, wherein the construction unit is configured to obtain an image data set, encode each image data included in the image data set into corresponding first quantum data; and train an initial quantum neural network using the first quantum data corresponding to each image data to obtain a quantum classifier.
- 根据权利要求8所述的装置,其中,所述构建单元,配置为将所述各图像数据对应的第一量 子数据输入至初始量子神经网络,得到第二量子数据;利用预设测量方法测量所述第二量子数据,得到所述第二量子数据对应的第一测量数据;对所述第一测量数据进行偏置处理得到位于预设映射区间的第二测量数据;基于所述第二测量数据以及所述第一量子数据确定所述初始量子神经网络的损失函数;确定所述损失函数是否满足第二条件;在所述损失函数不满足所述第二条件的情况下,重复执行如下步骤,直至所述损失函数满足所述第二条件:将所述各图像数据对应的第一量子数据输入至初始量子神经网络,得到第二量子数据;利用预设测量方法测量所述第二量子数据,得到所述第二量子数据对应的第一测量数据;对所述第一测量数据进行偏置处理得到位于预设映射区间的第二测量数据;基于所述第二测量数据以及所述第一量子数据确定所述初始量子神经网络的损失函数。The device according to claim 8, wherein the construction unit is configured to convert the first quantity corresponding to each image data into The sub-data is input into the initial quantum neural network to obtain the second quantum data; the second quantum data is measured by a preset measurement method to obtain the first measurement data corresponding to the second quantum data; the first measurement data is biased to obtain the second measurement data located in the preset mapping interval; the loss function of the initial quantum neural network is determined based on the second measurement data and the first quantum data; it is determined whether the loss function satisfies the second condition; if the loss function does not satisfy the second condition, the following steps are repeated until the loss function satisfies the second condition: the first quantum data corresponding to each image data is input into the initial quantum neural network to obtain the second quantum data; the second quantum data is measured by a preset measurement method to obtain the first measurement data corresponding to the second quantum data; the first measurement data is biased to obtain the second measurement data located in the preset mapping interval; the loss function of the initial quantum neural network is determined based on the second measurement data and the first quantum data.
- 根据权利要求9所述的装置,其中,所述构建单元,配置为在所述损失函数满足第二条件的情况下,将所述损失函数对应的所述初始量子神经网络的参数确定为目标参数,并基于所述目标参数得到所述量子分类器。The device according to claim 9, wherein the construction unit is configured to determine the parameters of the initial quantum neural network corresponding to the loss function as target parameters when the loss function satisfies the second condition, and obtain the quantum classifier based on the target parameters.
- 根据权利要求8至10中任一项所述的装置,其中,所述图像数据集中的各图像数据为N维数据,N为大于等于1的整数;所述构建单元,配置为将所述图像数据集中的各图像数据设置为预设形式的图像数据;基于所述各图像数据对应的预设形式的图像数据确定初始状态的量子态数据;基于所述各图像数据对应的预设形式的图像数据,将所述各图像数据编码为量子门形式并作用在所述初始状态的量子态数据上,得到所述各图像数据对应的第一量子数据。The device according to any one of claims 8 to 10, wherein each image data in the image data set is N-dimensional data, and N is an integer greater than or equal to 1; the construction unit is configured to set each image data in the image data set as image data in a preset form; determine the quantum state data of the initial state based on the image data in the preset form corresponding to each image data; based on the image data in the preset form corresponding to each image data, encode each image data into a quantum gate form and act on the quantum state data of the initial state to obtain the first quantum data corresponding to each image data.
- 根据权利要求7至10中任一项所述的装置,其中,所述确定单元,配置为利用所述量子分类器产生的决策边界确定所述多个待评价图像中的各待评价图像在所述决策边界中所处的区域,将位于所述决策边界的中心区域的主观评价值对应的待评价图像确定为所述多个待评价图像中图像质量满足第一条件的目标图像。The device according to any one of claims 7 to 10, wherein the determination unit is configured to use the decision boundary generated by the quantum classifier to determine the area where each image to be evaluated among the multiple images to be evaluated is located in the decision boundary, and determine the image to be evaluated corresponding to the subjective evaluation value located in the central area of the decision boundary as the target image whose image quality meets the first condition among the multiple images to be evaluated.
- 一种电子设备,所述电子设备包括:存储器和处理器,所述存储器上存储有计算机可执行指令,所述处理器运行所述存储器上的计算机可执行指令时可实现权利要求1至6中任一项所述的图像质量评价方法。An electronic device comprises: a memory and a processor, wherein the memory stores computer executable instructions, and when the processor runs the computer executable instructions on the memory, the image quality assessment method according to any one of claims 1 to 6 can be implemented.
- 一种计算机存储介质,所述存储介质上存储有可执行指令,该可执行指令被处理器执行时实现权利要求1至6中任一项所述的图像质量评价方法。A computer storage medium having executable instructions stored thereon, wherein the executable instructions, when executed by a processor, implement the image quality assessment method according to any one of claims 1 to 6.
- 一种计算机程序产品,包括计算机可读代码,在所述计算机可读代码在电子设备中运行的情况下,所述电子设备中的处理器执行如权利要求1至6中任一项所述的图像质量评价方法。 A computer program product comprises a computer readable code. When the computer readable code is executed in an electronic device, a processor in the electronic device executes the image quality assessment method according to any one of claims 1 to 6.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211252465.7 | 2022-10-13 | ||
CN202211252465.7A CN116797913A (en) | 2022-10-13 | 2022-10-13 | Image quality evaluation method and device, electronic equipment and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2024078394A1 true WO2024078394A1 (en) | 2024-04-18 |
Family
ID=88042553
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2023/123283 WO2024078394A1 (en) | 2022-10-13 | 2023-10-07 | Image quality evaluation method and apparatus, and electronic device, storage medium and program product |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN116797913A (en) |
WO (1) | WO2024078394A1 (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116797913A (en) * | 2022-10-13 | 2023-09-22 | 中移(苏州)软件技术有限公司 | Image quality evaluation method and device, electronic equipment and storage medium |
CN116993839B (en) * | 2023-09-26 | 2024-01-26 | 苏州元脑智能科技有限公司 | Coding mode screening method and device, electronic equipment and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20200005154A1 (en) * | 2018-02-01 | 2020-01-02 | Siemens Healthcare Limited | Data encoding and classification |
CN114419380A (en) * | 2022-03-31 | 2022-04-29 | 苏州浪潮智能科技有限公司 | Image identification method and device, electronic equipment and storage medium |
CN114792378A (en) * | 2021-03-30 | 2022-07-26 | 合肥本源量子计算科技有限责任公司 | Quantum image identification method and device |
CN116797913A (en) * | 2022-10-13 | 2023-09-22 | 中移(苏州)软件技术有限公司 | Image quality evaluation method and device, electronic equipment and storage medium |
-
2022
- 2022-10-13 CN CN202211252465.7A patent/CN116797913A/en active Pending
-
2023
- 2023-10-07 WO PCT/CN2023/123283 patent/WO2024078394A1/en unknown
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20200005154A1 (en) * | 2018-02-01 | 2020-01-02 | Siemens Healthcare Limited | Data encoding and classification |
CN114792378A (en) * | 2021-03-30 | 2022-07-26 | 合肥本源量子计算科技有限责任公司 | Quantum image identification method and device |
CN114419380A (en) * | 2022-03-31 | 2022-04-29 | 苏州浪潮智能科技有限公司 | Image identification method and device, electronic equipment and storage medium |
CN116797913A (en) * | 2022-10-13 | 2023-09-22 | 中移(苏州)软件技术有限公司 | Image quality evaluation method and device, electronic equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN116797913A (en) | 2023-09-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2024078394A1 (en) | Image quality evaluation method and apparatus, and electronic device, storage medium and program product | |
CN111461168B (en) | Training sample expansion method and device, electronic equipment and storage medium | |
Zhang et al. | Impact of imagenet model selection on domain adaptation | |
CN112117011A (en) | Infectious disease early risk early warning method and device based on artificial intelligence | |
WO2022237153A1 (en) | Target detection method and model training method therefor, related apparatus, medium, and program product | |
JPWO2019073923A1 (en) | Abnormal product judgment method | |
WO2021132099A1 (en) | Learning support device, learning device, learning support method, and learning support program | |
CN115051929B (en) | Network fault prediction method and device based on self-supervision target perception neural network | |
US20220301106A1 (en) | Training method and apparatus for image processing model, and image processing method and apparatus | |
KR20230165085A (en) | Method and system for quantitatively evaluating alignment between multimodal feature vectors | |
CN115471885A (en) | Action unit correlation learning method and device, electronic device and storage medium | |
CN113486925B (en) | Model training method, fundus image generation method, model evaluation method and device | |
CN111651626B (en) | Image classification method, device and readable storage medium | |
WO2021114637A1 (en) | Deep neural network-based method and device for data processing | |
CN108509732A (en) | Appraisal procedure, terminal device and the storage medium of steam turbine fault severity level | |
CN116384502B (en) | Method, device, equipment and medium for calculating contribution of participant value in federal learning | |
CN114328174A (en) | Multi-view software defect prediction method and system based on counterstudy | |
WO2021114626A1 (en) | Method for detecting quality of medical record data and related device | |
CN113627538A (en) | Method and electronic device for training asymmetric generation countermeasure network to generate image | |
WO2021038840A1 (en) | Object number estimation device, control method, and program | |
CN114186078B (en) | Knowledge graph representation learning method and device, storage medium and electronic equipment | |
WO2024108396A1 (en) | Method and apparatus for measuring hemodynamic heterogeneity and computer device | |
WO2022070256A1 (en) | Information processing device, information processing method, and recording medium | |
CN110738638A (en) | Visual saliency detection algorithm applicability prediction and performance blind evaluation method | |
JP7309134B2 (en) | Visualization method, program used therefor, visualization device, and discriminating device provided with the same |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 23876596 Country of ref document: EP Kind code of ref document: A1 |