CN109977815A - Image quality evaluating method and device, electronic equipment, storage medium - Google Patents
Image quality evaluating method and device, electronic equipment, storage medium Download PDFInfo
- Publication number
- CN109977815A CN109977815A CN201910189151.9A CN201910189151A CN109977815A CN 109977815 A CN109977815 A CN 109977815A CN 201910189151 A CN201910189151 A CN 201910189151A CN 109977815 A CN109977815 A CN 109977815A
- Authority
- CN
- China
- Prior art keywords
- image quality
- quality evaluation
- face
- trained
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 49
- 230000001815 facial effect Effects 0.000 claims abstract description 102
- 239000013598 vector Substances 0.000 claims abstract description 87
- 238000012549 training Methods 0.000 claims abstract description 73
- 238000011156 evaluation Methods 0.000 claims abstract description 71
- 238000013441 quality evaluation Methods 0.000 claims description 220
- 230000015654 memory Effects 0.000 claims description 20
- 238000004891 communication Methods 0.000 claims description 9
- 230000007613 environmental effect Effects 0.000 abstract description 2
- 238000013210 evaluation model Methods 0.000 abstract 2
- 238000010586 diagram Methods 0.000 description 17
- 238000000605 extraction Methods 0.000 description 11
- 238000005259 measurement Methods 0.000 description 10
- 238000004590 computer program Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 238000001514 detection method Methods 0.000 description 4
- 239000000284 extract Substances 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the present disclosure discloses a kind of image quality evaluating method, this method comprises: obtaining at least one face characteristic parameter of facial image to be evaluated；According at least one face characteristic parameter, face feature vector to be evaluated is obtained；Based on facial image Environmental Evaluation Model has been trained, the corresponding quality of human face image evaluation result of face feature vector to be evaluated is obtained.The technical solution of the embodiment of the present disclosure, by obtaining the corresponding face feature vector of facial image, to combine based on objective measure and the subjective quality of human face image evaluation model measuring training and obtaining, the evaluation of quality of human face image is carried out, to improve the accuracy of evaluation result.
Description
Technical Field
The present disclosure relates to the field of computer vision technologies, and in particular, to an image quality evaluation method and apparatus, an electronic device, and a storage medium.
Background
With the continuous development of science and technology, face recognition technology is widely applied to various industries, for example, the face recognition technology is used for recognizing people in surveillance videos. In the face recognition technology, if the provided face image has the conditions of blurring, serious shielding and distortion, namely the quality of the face image is poor, the accuracy of recognition is directly affected, and the computing resources required by recognition are wasted.
From the aspect of performance and accuracy, the existing face image quality evaluation methods are mainly divided into traditional feature point evaluation and a face image evaluation method based on deep learning, and although the traditional feature point evaluation mechanism is high in operation speed, the traditional feature point evaluation mechanism is poor in environmental robustness and low in accuracy.
Disclosure of Invention
The embodiment of the disclosure is expected to provide an image quality evaluation method and device, an electronic device and a storage medium, wherein the evaluation of the quality of a face image is performed by acquiring a face feature vector corresponding to the face image and combining a face image quality evaluation model obtained based on objective measurement and subjective measurement training, so that the accuracy of an evaluation result is improved.
The technical scheme of the embodiment of the disclosure is realized as follows:
the embodiment of the disclosure provides an image quality evaluation method, which comprises the following steps:
acquiring at least one face characteristic parameter of a face image to be evaluated;
obtaining a face feature vector to be evaluated according to the at least one face feature parameter;
and acquiring a face image quality evaluation result corresponding to the face feature vector to be evaluated based on the trained face image quality evaluation model.
In the above scheme, before the obtaining, based on the trained facial image quality evaluation model, a facial image quality evaluation result corresponding to the facial feature vector to be evaluated, the method further includes:
generating the trained facial image quality evaluation model;
the generation process of the trained facial image quality evaluation model comprises the following steps:
acquiring a sample image, a quality subjective evaluation result thereof and a sample face feature vector;
and training a facial image quality evaluation model to be trained according to the quality subjective evaluation result of the sample image and the sample facial feature vector to obtain the trained facial image quality evaluation model.
In the above scheme, the generating process of the trained facial image quality evaluation model includes:
obtaining at least one sample characteristic parameter of the sample image;
and combining the at least one sample characteristic parameter to obtain the sample face characteristic vector.
In the above scheme, the generating process of the trained facial image quality evaluation model includes:
acquiring preset training parameters of the facial image quality evaluation model to be trained;
constructing a model training target condition according to the preset training parameters;
and training the facial image quality evaluation model to be trained based on the model training target condition according to the quality subjective evaluation result and the sample facial feature vector to obtain the trained facial image quality evaluation model.
In the above scheme, the training of the to-be-trained face image quality evaluation model includes:
and performing fitting regression on the sample facial feature vectors and the quality subjective evaluation results to obtain the trained facial image quality evaluation model meeting the model training target conditions.
In the above scheme, the quality evaluation model of the face image to be trained is:
yn＝WTZn+b
wherein W and b are parameters of a model to be trained in the human face image quality evaluation model to be trained, and ZnTo input data, ynTo output data.
In the above scheme, the model training target conditions are:
wherein C and epsilon are the preset training parameters.
In the above scheme, after obtaining the quality evaluation result of the face image corresponding to the to-be-evaluated face feature vector, the method includes:
determining whether the human face image quality evaluation result meets the requirement of a preset image quality result;
and under the condition that the quality evaluation result of the face image meets the preset image quality result requirement, determining the face image to be evaluated as a target image for face recognition.
The embodiment of the present disclosure provides an image quality evaluation device, including:
the acquisition module is used for acquiring at least one face characteristic parameter of a face image to be evaluated; obtaining a face feature vector to be evaluated according to the at least one face feature parameter;
and the evaluation module is used for acquiring a face image quality evaluation result corresponding to the face feature vector to be evaluated based on the trained face image quality evaluation model.
In the above image quality evaluation apparatus, the image quality evaluation apparatus further includes: a training module;
the training module is used for generating the trained facial image quality evaluation model;
the acquisition module is also used for acquiring a sample image, a quality subjective evaluation result thereof and a sample face feature vector;
the training module is specifically used for training a facial image quality evaluation model to be trained according to the quality subjective evaluation result of the sample image and the sample facial feature vector, and obtaining the trained facial image quality evaluation model.
In the image quality evaluation device, the obtaining module is specifically configured to obtain at least one sample characteristic parameter of the sample face image; and combining the at least one sample characteristic parameter to obtain the sample face characteristic vector.
In the image quality evaluation device, the training module is specifically configured to acquire preset training parameters of the facial image quality evaluation model to be trained; constructing a model training target condition according to the preset training parameters; and training the facial image quality evaluation model to be trained based on the model training target condition according to the quality subjective evaluation result and the sample facial feature vector to obtain the trained facial image quality evaluation model.
In the image quality evaluation device, the training module is specifically configured to perform fitting regression on the sample facial feature vectors and the quality subjective evaluation results to obtain the trained facial image quality evaluation model satisfying the model training target condition.
In the above image quality evaluation apparatus, the facial image quality evaluation model to be trained is:
yn＝WTZn+b
wherein W and b are parameters of a model to be trained in the human face image quality evaluation model to be trained, and ZnTo input data, ynTo output data.
In the image quality evaluation device, the model training target conditions are:
wherein C and epsilon are the preset training parameters.
In the image quality evaluation device, the evaluation module is further configured to determine whether the face image quality evaluation result meets a preset image quality result requirement; and under the condition that the quality evaluation result of the face image meets the preset image quality result requirement, determining the face image to be evaluated as a target image for face recognition.
An embodiment of the present disclosure provides an electronic device, including: a processor, a memory, and a communication bus; wherein,
the communication bus is used for realizing connection communication between the processor and the memory;
the processor is used for executing the image quality evaluation program stored in the memory so as to realize the image quality evaluation method.
The electronic equipment is a mobile phone, a tablet computer, a server or edge node equipment.
The disclosed embodiments provide a computer-readable storage medium storing one or more programs, which may be executed by one or more processors, to implement the above-described image quality evaluation method.
The image quality evaluation method provided by the embodiment of the disclosure extracts at least one face characteristic parameter of a face image to be evaluated; combining according to at least one face characteristic parameter to obtain a face characteristic vector to be evaluated; and acquiring a face image quality evaluation result corresponding to the face feature vector to be evaluated based on the trained face image quality evaluation model. That is to say, according to the technical scheme provided by the embodiment of the disclosure, the facial feature vector corresponding to the facial image is obtained, and the facial image quality evaluation model obtained based on objective measurement and subjective measurement training is combined to evaluate the facial image quality, so that the accuracy of the evaluation result is improved.
Drawings
Fig. 1 is a first schematic flow chart of an image quality evaluation method according to an embodiment of the present disclosure;
fig. 2 is a schematic diagram of exemplary obtaining of a feature vector of a face to be evaluated according to an embodiment of the present disclosure;
fig. 3 is a schematic diagram illustrating an exemplary determination of a result of evaluating the quality of a face image according to an embodiment of the present disclosure;
fig. 4 is a schematic flow chart diagram of an image quality evaluation method according to an embodiment of the present disclosure;
fig. 5 is a schematic structural diagram of an image quality evaluation apparatus provided in an embodiment of the present disclosure;
fig. 6 is a schematic structural diagram of an electronic device according to an embodiment of the present disclosure.
Detailed Description
The technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the drawings in the embodiments of the present disclosure.
The embodiment of the disclosure provides an image quality evaluation method. Fig. 1 is a first schematic flow chart of an image quality evaluation method according to an embodiment of the present disclosure. As shown in fig. 1, the method mainly comprises the following steps:
s101, at least one face characteristic parameter of a face image to be evaluated is obtained.
In the embodiment of the disclosure, the image quality evaluation device may perform feature parameter extraction processing on the face image to be evaluated, and obtain at least one corresponding face feature parameter from the face image to be evaluated.
It should be noted that, in the embodiment of the present disclosure, the face image to be evaluated is any face image that needs to be subjected to quality evaluation, and the specific face image to be evaluated is not limited in the embodiment of the present disclosure. The image quality evaluation device stores a preset feature extraction method for extracting at least one face feature parameter of any one face image. The preset feature extraction method may be a neural network, a principal component analysis algorithm, a laplacian feature map method, and the like, which can extract the face features, and the specific extraction process is the prior art and is not described herein again, and the specific preset feature extraction method is not limited in the embodiments of the present disclosure. The face feature parameters may include: the face feature parameters include, but are not limited to, face yaw angle, pitch angle, roll angle, eye distance, face size, motion speed, detection score, keypoint score, and blur score.
Illustratively, in the embodiment of the disclosure, the image quality evaluation device extracts, for the face image to be evaluated, a corresponding face yaw angle, a pitch angle, a roll angle, an eye distance, a face size, a motion speed, a detection score, a key point score, and a blur score according to a preset feature extraction method, and these data are face feature parameters to be evaluated corresponding to the image to be evaluated.
And S102, obtaining a face feature vector to be evaluated according to at least one face feature parameter.
In the embodiment of the disclosure, after the image quality evaluation device obtains at least one face feature parameter of the face image to be evaluated, a face feature vector to be evaluated can be obtained according to the at least one face feature parameter.
It should be noted that, in the embodiment of the present disclosure, the image quality evaluation device may combine at least one face feature parameter according to a preset sequence, so as to obtain a face feature vector to be evaluated, where the specific preset sequence is not limited in the embodiment of the present disclosure.
For example, fig. 2 is a schematic diagram of exemplary obtaining of a feature vector of a face to be evaluated according to an embodiment of the present disclosure. As shown in fig. 2, the image quality evaluation device combines the face feature parameters according to a face yaw angle a1, a pitch angle a2, a roll angle A3, an eye distance a4, a face size a5, a motion speed A6, a detection score a7, a key point score A8, and a blur score a9, which are used to acquire a face image to be evaluated, in a preset order, so as to obtain a face feature vector to be evaluated (a5, a1, a4, a2, A3, A6, a7, A8, a 9).
It should be noted that, in the embodiment of the present disclosure, a preset sequence is stored in the image quality evaluation apparatus, and is used to combine at least one facial feature parameter of a facial image to be evaluated, so as to serve as a basis for evaluating the quality of the facial image to be evaluated, and input the combined facial feature parameter into a trained facial image quality evaluation model, where each type of facial feature parameter has a corresponding weight coefficient in the model, and therefore, at least one facial feature parameter may be combined according to the weight coefficient corresponding to each type of facial feature in the model.
S103, obtaining a face image quality evaluation result corresponding to the face feature vector to be evaluated based on the trained face image quality evaluation model.
In the embodiment of the disclosure, after the image quality evaluation device obtains the facial feature vector to be evaluated, a facial image quality evaluation result corresponding to the facial feature vector to be evaluated can be obtained based on the trained facial image quality evaluation model.
It should be noted that, in the embodiment of the present disclosure, the trained facial image quality evaluation model is a facial image quality evaluation model obtained by an image quality evaluation device based on objective measurement and subjective measurement training.
Specifically, in the embodiment of the present disclosure, the image quality evaluation device may input the feature vector of the face to be evaluated into the trained face image quality evaluation model, and the trained face image quality evaluation model outputs the face image quality evaluation result, where the face image quality evaluation result represents the quality of the face image to be evaluated.
It should be noted that, in the embodiment of the present disclosure, before obtaining a facial image quality evaluation result corresponding to a facial feature vector to be evaluated based on a trained facial image quality evaluation model, an image quality evaluation device needs to generate the trained facial image quality evaluation model, and mainly includes the following steps: acquiring a sample image, a quality subjective evaluation result thereof and a sample face feature vector; and training a facial image quality evaluation model to be trained according to the quality subjective evaluation result of the sample image and the sample facial feature vector to obtain the trained facial image quality evaluation model.
The image quality evaluation device combines the subjective quality evaluation result of the sample image and the sample face feature vector, namely subjective evaluation and objective evaluation, by using the face image quality evaluation model to be trained to obtain the trained face image quality evaluation model, so that the evaluation result obtained by applying the trained face image quality evaluation model to carry out the face image quality evaluation meets both the computer subjective evaluation and the human subjective evaluation. In addition, a deep neural network based on deep learning is usually adopted for evaluating the quality of the face image, however, the deep neural network is difficult to deploy in an edge node with limited computing capability, and in the embodiment of the disclosure, the image quality evaluation device utilizes a traditional feature extraction algorithm in combination with a trained face image quality evaluation model, actually determines the quality of the face image according to an image feature vector mechanism, and is lighter than the deep learning-based mechanism, so that the image quality evaluation device is suitable for deploying in the edge node, can perform rapid calculation in the edge node, and reduces the calculation burden of the edge node.
It should be noted that, in the embodiments of the present disclosure, a plurality of sample face images may be included, for example, different face images acquired in different scenes or different environments may be used as the sample images, specific sample images, and the number of the sample images is not limited in the embodiments of the present disclosure.
Specifically, in the embodiment of the present disclosure, a manual marking mode may be adopted to manually score the sample image, so as to obtain a quality subjective evaluation result of the sample image. Then, the quality subjective evaluation result is input into an image quality evaluation device, and the image quality evaluation device can obtain the quality subjective evaluation result of the sample face image.
For example, each sample image is manually scored, and the score is determined as the corresponding quality subjective evaluation result. For example, the score range may be limited to 0-100 points, i.e., if a certain sample is subjectively considered to be of good image quality, a higher score may be scored, whereas a lower score may be scored.
Illustratively, in the embodiment of the present disclosure, there are 5 sample images, which are: the image quality evaluation device can acquire a sample image 1 with a quality subjective evaluation result of 70 points, a sample image 2 with a quality subjective evaluation result of 60 points, a sample image 3 with a quality subjective evaluation result of 90 points, a sample image 4 with a quality subjective evaluation result of 55 points and a sample face image 5 with a quality subjective evaluation result of 85 points.
Specifically, in an embodiment of the present disclosure, an image quality evaluation apparatus obtains a sample feature vector of a sample image, including: obtaining at least one sample characteristic parameter of a sample image; and combining at least one sample characteristic parameter to obtain a sample face characteristic vector.
It should be noted that, in the embodiment of the present disclosure, the image quality evaluation device performs feature extraction and feature combination on the sample image, so that the process of obtaining the sample feature parameters is completely consistent with the manner of performing feature extraction and feature combination on the face image to be evaluated in steps S101 and S102 to obtain the face feature vector to be evaluated, and details are not repeated here.
In the embodiment of the disclosure, after the image quality evaluation device obtains the quality subjective evaluation result and the sample face feature vector, the image quality evaluation device can train the face image quality evaluation model to be trained according to the quality subjective evaluation result and the sample face feature vector to obtain the trained face image quality evaluation model.
Specifically, in the embodiment of the present disclosure, the quality evaluation model of the face image to be trained is shown in formula (1):
yn＝WTZn+b (1)
wherein Z isnFor input data, i.e. face feature vectors, y, corresponding to arbitrary face imagesnIn order to output data, namely the determined quality evaluation result of the face image, W and b are parameters of a model to be trained, and adjustment is needed in the training process.
It should be noted that, in the embodiment of the present disclosure, the image quality evaluation device trains an initial human face image quality evaluation model to be trained according to a quality subjective evaluation result and a sample human face feature Vector, the human face image quality evaluation model to be trained may be a Support Vector Regression (SVR) type model shown in formula (1), the model allows an error within a certain range, and only if the error exceeds the certain range, the model calculates a loss, and is suitable for an application scenario with a small data volume.
Specifically, in the embodiment of the present disclosure, the image quality evaluation device trains a facial image quality evaluation model to be trained according to a quality subjective evaluation result and a sample facial feature vector, and obtains the trained facial image quality evaluation model, including: acquiring preset training parameters of a facial image quality evaluation model to be trained; constructing a model training target condition according to a preset training parameter; and training the facial image quality evaluation model to be trained based on the model training target condition according to the quality subjective evaluation result and the sample facial feature vector to obtain the trained facial image quality evaluation model.
It can be understood that, in the embodiment of the present disclosure, the image quality evaluation device trains the facial image quality evaluation model to be trained, and the purpose is to train the model to output an evaluation result that better meets the artificial subjective evaluation according to the obtained objective parameters, that is, the facial feature vectors corresponding to the facial images, so that a model training target condition needs to be constructed for constraint, so that the finally obtained trained facial image quality evaluation result can achieve the purpose.
Specifically, in the embodiment of the present disclosure, the model training target condition constructed by the image quality evaluation apparatus is as shown in formula (2):
c and epsilon are preset training parameters of a model supporting human face image quality evaluation to be trained, and are stored in an image quality evaluation device in advance, and ZnFor the input data, i.e. the face feature vector corresponding to the face image, in the training process, the input data is actually the sample face feature vector, ynSpecific preset training parameters are subjective evaluation results of the quality of the sample images, and the embodiment of the disclosure is not limited.
Specifically, in the embodiment of the present disclosure, the training of the image quality evaluation device on the human face image quality evaluation model to be trained includes: and performing fitting regression on the sample facial feature vectors and the quality subjective evaluation results to obtain a trained facial image quality evaluation model meeting the model training target conditions.
It should be noted that, in the embodiment of the present disclosure, N sample images may be included, and accordinglyThe method can include N sample face feature vectors, where N is a natural number greater than or equal to 1, and in the training process, as shown in formula (2), the face feature vectors substituted by the image quality evaluation device, that is, N sample face feature vectors, that is, Z sample face feature vectors1、Z2、……，ZNCorresponding, y1、y2、……，yNAnd adjusting the values of W and b in the calculation process of respectively and correspondingly substituting the values into the formula (2) until the formula (2) reaches the minimum value, wherein the values of W and b at the moment are the specific numerical values required by the final formula (1), so as to obtain the trained facial image quality evaluation model.
Fig. 3 is a schematic diagram illustrating an exemplary determination of a result of evaluating the quality of a face image according to an embodiment of the present disclosure. As shown in fig. 3, the image quality evaluation apparatus extracts a face yaw angle a1, a pitch angle a2, a roll angle A3, an eye distance A4, a face size A5, a movement speed A6, a detection score a7, a keypoint score A8, and a blur score a9 corresponding to a face image to be evaluated according to a preset feature extraction method, and then combines a1, a2, A3, A4, A5, A6, a7, A8, and a9 according to a preset sequence to obtain a face feature vector to be evaluated (A5, a1, A4, a2, A3, A6, a7, A8, a9), and inputs the face feature vector to be evaluated into a trained face image quality evaluation model, so as to obtain a corresponding face image quality evaluation result.
Fig. 4 is a schematic flowchart of a second image quality evaluation method according to an embodiment of the present disclosure. As shown in fig. 4, after obtaining the face image quality evaluation result, the image quality evaluation apparatus may further perform the following steps:
and S104, determining whether the human face image quality evaluation result meets the preset image quality result requirement.
In the embodiment of the disclosure, after obtaining the face image quality evaluation result corresponding to the face feature vector to be evaluated, the image quality evaluation device may further determine whether the face image quality evaluation result meets the preset image quality result requirement.
It should be noted that, in the embodiment of the present disclosure, the image quality evaluation device may store a preset image quality result requirement, which may be predetermined according to an actual requirement of a user or an actual recognition capability of a face recognition technology. The specific preset image quality result requires that the embodiment of the present disclosure is not limited.
For example, in the embodiment of the present disclosure, the preset image quality result is required to be that the face image quality evaluation result is greater than 80 minutes, therefore, if the face image evaluation result corresponding to the to-be-evaluated face feature vector is greater than 80 minutes, the image quality evaluation device determines that the to-be-evaluated face image meets the preset image quality result requirement, and correspondingly, if the face image evaluation result corresponding to the to-be-evaluated face feature vector is less than or equal to 80 minutes, the image quality evaluation device determines that the to-be-evaluated face image does not meet the preset image quality result requirement.
And S105, determining the face image to be evaluated as a target image for face recognition under the condition that the face image quality evaluation result meets the preset image quality result requirement.
In the embodiment of the disclosure, the image quality evaluation device further determines the face image to be evaluated as a target image for face recognition when the face image quality evaluation result meets the preset image quality result requirement.
It can be understood that, in the embodiment of the present disclosure, if the face image quality evaluation result meets the preset image quality result requirement, it indicates that the face image to be evaluated performs better in all aspects, and therefore, the face image to be evaluated can be used for face recognition, and if the face image quality evaluation result does not meet the preset image quality result requirement, it indicates that the comprehensive aspects of the face image to be evaluated are difficult to support face recognition, and the face image meeting the preset image quality result requirement is determined as the target image for face recognition, so that the face image with poor image quality can be prevented from being recognized, and the calculation resources required by face recognition are saved.
It can be understood that, in the embodiment of the present disclosure, an image objective measurement method and a trained facial image quality evaluation model performing linear fitting based on subjective evaluation and objective evaluation are adopted to perform quality evaluation on a facial image to be evaluated, so that the problem of deviation between subjective evaluation and subjective evaluation in a facial image quality evaluation mechanism is solved, and an accurate image quality evaluation result is obtained.
The embodiment of the disclosure provides an image quality evaluation method, which includes the steps of obtaining at least one face characteristic parameter of a face image to be evaluated; obtaining a face feature vector to be evaluated according to at least one face feature parameter; and acquiring a face image quality evaluation result corresponding to the face feature vector to be evaluated based on the trained face image quality evaluation model. That is to say, according to the technical scheme provided by the embodiment of the disclosure, the facial feature vector corresponding to the facial image is obtained, and the facial image quality evaluation model obtained based on objective measurement and subjective measurement training is combined to evaluate the facial image quality, so that the accuracy of the evaluation result is improved.
The embodiment of the disclosure provides an image quality evaluation device. Fig. 5 is a schematic structural diagram of an image quality evaluation apparatus according to an embodiment of the present disclosure. As shown in fig. 5, the image quality evaluation apparatus includes:
an obtaining module 501, configured to obtain at least one face feature parameter of a face image to be evaluated; obtaining a face feature vector to be evaluated according to the at least one face feature parameter;
the evaluation module 502 is configured to obtain a face image quality evaluation result corresponding to the to-be-evaluated face feature vector based on the trained face image quality evaluation model.
Optionally, the image quality evaluation apparatus further includes: a training module 503;
the training module 503 is configured to generate the trained facial image quality evaluation model;
the obtaining module 501 is further configured to obtain a sample image, a quality subjective evaluation result thereof, and a sample face feature vector;
the training module 503 is specifically configured to train a facial image quality evaluation model to be trained according to the quality subjective evaluation result of the sample image and the sample facial feature vector, and obtain the trained facial image quality evaluation model.
It can be understood that the training module 503 combines the subjective quality evaluation result of the sample image and the sample face feature vector, i.e. subjective evaluation and objective evaluation, with the face image quality evaluation model to be trained to obtain the trained face image quality evaluation model, so that the evaluation result obtained by applying the trained face image quality evaluation model to perform the face image quality evaluation meets both the computer subjective evaluation and the human subjective evaluation. In addition, a deep neural network based on deep learning is usually adopted for evaluating the quality of the face image, however, the deep neural network is difficult to deploy in an edge node with limited computing capability, and in the embodiment of the disclosure, the image quality evaluation device utilizes a traditional feature extraction algorithm in combination with a trained face image quality evaluation model, actually determines the quality of the face image according to an image feature vector mechanism, and is lighter than the deep learning-based mechanism, so that the image quality evaluation device is suitable for deploying in the edge node, can perform rapid calculation in the edge node, and reduces the calculation burden of the edge node.
Optionally, the obtaining module 501 is specifically configured to obtain at least one sample feature parameter of the sample face image; and combining the at least one sample characteristic parameter to obtain the sample face characteristic vector.
Optionally, the training module 503 is specifically configured to acquire a preset training parameter of the facial image quality evaluation model to be trained; constructing a model training target condition according to the preset training parameters; and training the facial image quality evaluation model to be trained based on the model training target condition according to the quality subjective evaluation result and the sample facial feature vector to obtain the trained facial image quality evaluation model.
Optionally, the training module 503 is specifically configured to perform fitting regression on the sample facial feature vectors and the quality subjective evaluation results to obtain the trained facial image quality evaluation model meeting the model training target condition.
Optionally, the quality evaluation model of the face image to be trained is as follows:
yn＝WTZn+b
wherein W and b are parameters of a model to be trained in the human face image quality evaluation model to be trained, and ZnTo input data, ynTo output data.
It should be noted that, in the embodiment of the present disclosure, the training module 503 is a training face image quality evaluation model to be initially trained according to the quality subjective evaluation result and the sample face feature vector, the training face image quality evaluation model to be trained may be an SVR type model, the model allows an error within a certain range, and only if the error exceeds a certain range, the model calculates a loss, and is suitable for an application scenario with a small data size, of course, the training face image quality evaluation model may also be other models such as a convolutional neural network, and a specific face image quality evaluation model to be trained is not limited in the embodiment of the present disclosure.
Optionally, the model training target conditions are:
wherein C and epsilon are the preset training parameters.
Optionally, the evaluation module 503 is further configured to determine whether the face image quality evaluation result meets a preset image quality result requirement; and under the condition that the quality evaluation result of the face image meets the preset image quality result requirement, determining the face image to be evaluated as a target image for face recognition.
The embodiment of the disclosure provides an image quality evaluation device, which acquires at least one face characteristic parameter of a face image to be evaluated; obtaining a face feature vector to be evaluated according to at least one face feature parameter; and acquiring a face image quality evaluation result corresponding to the face feature vector to be evaluated based on the trained face image quality evaluation model. That is to say, the image quality evaluation device provided by the embodiment of the present disclosure performs the evaluation of the quality of the face image by acquiring the face feature vector corresponding to the face image and combining the face image quality evaluation model obtained based on objective measurement and subjective measurement training, thereby improving the accuracy of the evaluation result.
The embodiment of the disclosure also provides an electronic device. Fig. 6 is a schematic structural diagram of an electronic device according to an embodiment of the present disclosure. As shown in fig. 6, the electronic apparatus includes: a processor 601, memory 602, and a communication bus 603; wherein,
the communication bus 603 is used for realizing connection communication between the processor 601 and the memory 602;
the processor 601 is configured to execute the image quality evaluation program stored in the memory 602 to implement the image quality evaluation method.
Optionally, the electronic device is a mobile phone, a tablet computer, a server, or an edge node device. Of course, the electronic device may be other types or forms of devices as long as the above-described image quality evaluation method can be performed, and the specific electronic device is not limited in the embodiment of the present disclosure.
The disclosed embodiments also provide a computer-readable storage medium storing one or more programs, which can be executed by one or more processors to implement the above-described image quality evaluation method. The computer-readable storage medium may be a volatile Memory (volatile Memory), such as a Random-Access Memory (RAM); or a non-volatile Memory (non-volatile Memory) such as a Read-Only Memory (ROM), a flash Memory (flash Memory), a Hard Disk (HDD) or a Solid-State Drive (SSD); or may be a respective device, such as a mobile phone, computer, tablet device, personal digital assistant, etc., that includes one or any combination of the above-mentioned memories.
As will be appreciated by one skilled in the art, embodiments of the present disclosure may be provided as a method, system, or computer program product. Accordingly, the present disclosure may take the form of a hardware embodiment, a software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present disclosure may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, and the like) having computer-usable program code embodied therein.
The present disclosure is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the disclosure. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable signal processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable signal processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable signal processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable signal processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above description is only for the preferred embodiment of the present disclosure, and is not intended to limit the scope of the present disclosure.
Claims (10)
1. An image quality evaluation method, characterized by comprising:
acquiring at least one face characteristic parameter of a face image to be evaluated;
obtaining a face feature vector to be evaluated according to the at least one face feature parameter;
and acquiring a face image quality evaluation result corresponding to the face feature vector to be evaluated based on the trained face image quality evaluation model.
2. The image quality evaluation method according to claim 1, wherein before obtaining the facial image quality evaluation result corresponding to the facial feature vector to be evaluated based on the trained facial image quality evaluation model, the method further comprises:
generating the trained facial image quality evaluation model;
the generation process of the trained facial image quality evaluation model comprises the following steps:
acquiring a sample image, a quality subjective evaluation result thereof and a sample face feature vector;
and training a facial image quality evaluation model to be trained according to the quality subjective evaluation result of the sample image and the sample facial feature vector to obtain the trained facial image quality evaluation model.
3. The image quality evaluation method according to claim 2, wherein the generation process of the trained human face image quality evaluation model comprises:
obtaining at least one sample characteristic parameter of the sample image;
and combining the at least one sample characteristic parameter to obtain the sample face characteristic vector.
4. The image quality evaluation method according to claim 2, wherein the generation process of the trained human face image quality evaluation model comprises:
acquiring preset training parameters of the facial image quality evaluation model to be trained;
constructing a model training target condition according to the preset training parameters;
and training the facial image quality evaluation model to be trained based on the model training target condition according to the quality subjective evaluation result and the sample facial feature vector to obtain the trained facial image quality evaluation model.
5. The image quality evaluation method according to claim 4, wherein the training of the facial image quality evaluation model to be trained comprises:
and performing fitting regression on the sample facial feature vectors and the quality subjective evaluation results to obtain the trained facial image quality evaluation model meeting the model training target conditions.
6. The image quality evaluation method according to claim 5, wherein the facial image quality evaluation model to be trained is:
yn＝WTZn+b
wherein W and b are parameters of a model to be trained in the human face image quality evaluation model to be trained, and ZnTo input data, ynTo output data.
7. The image quality evaluation method according to claim 6, wherein the model training target conditions are:
wherein C and epsilon are the preset training parameters.
8. An image quality evaluation apparatus, comprising:
the acquisition module is used for acquiring at least one face characteristic parameter of a face image to be evaluated; obtaining a face feature vector to be evaluated according to the at least one face feature parameter;
and the evaluation module is used for acquiring a face image quality evaluation result corresponding to the face feature vector to be evaluated based on the trained face image quality evaluation model.
9. An electronic device, characterized in that the electronic device comprises: a processor, a memory, and a communication bus; wherein,
the communication bus is used for realizing connection communication between the processor and the memory;
the processor is configured to execute the image quality evaluation program stored in the memory to implement the image quality evaluation method according to any one of claims 1 to 7.
10. A computer-readable storage medium storing one or more programs, the one or more programs being executable by one or more processors to implement the image quality evaluation method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910189151.9A CN109977815B (en) | 2019-03-13 | 2019-03-13 | Image quality evaluation method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910189151.9A CN109977815B (en) | 2019-03-13 | 2019-03-13 | Image quality evaluation method and device, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109977815A true CN109977815A (en) | 2019-07-05 |
| CN109977815B CN109977815B (en) | 2024-07-12 |
Family
ID=67078772
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910189151.9A Active CN109977815B (en) | 2019-03-13 | 2019-03-13 | Image quality evaluation method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109977815B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111881732A (en) * | 2020-06-17 | 2020-11-03 | 北京邮电大学 | SVM (support vector machine) -based face quality evaluation method |
| CN112597909A (en) * | 2020-12-25 | 2021-04-02 | 北京芯翌智能信息技术有限公司 | Method and equipment for evaluating quality of face picture |
| CN113012089A (en) * | 2019-12-19 | 2021-06-22 | 北京金山云网络技术有限公司 | Image quality evaluation method and device |
| CN113553971A (en) * | 2021-07-29 | 2021-10-26 | 青岛以萨数据技术有限公司 | Method and device for extracting optimal frame of face sequence and storage medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060147107A1 (en) * | 2005-01-04 | 2006-07-06 | Microsoft Corporation | Method and system for learning-based quality assessment of images |
| CN102201061A (en) * | 2011-06-24 | 2011-09-28 | 常州锐驰电子科技有限公司 | Intelligent safety monitoring system and method based on multilevel filtering face recognition |
| CN103824068A (en) * | 2014-03-19 | 2014-05-28 | 上海看看智能科技有限公司 | Human face payment authentication system and method |
| DE102014108924A1 (en) * | 2013-06-28 | 2015-01-15 | Cognex Corporation | A semi-supervised procedure for training an auxiliary model to recognize and capture multiple patterns |
| CN105979253A (en) * | 2016-05-06 | 2016-09-28 | 浙江科技学院 | Generalized regression neural network based non-reference stereoscopic image quality evaluation method |
| CN107292331A (en) * | 2017-05-22 | 2017-10-24 | 浙江科技学院 | Based on unsupervised feature learning without with reference to screen image quality evaluating method |
| CN108269250A (en) * | 2017-12-27 | 2018-07-10 | 武汉烽火众智数字技术有限责任公司 | Method and apparatus based on convolutional neural networks assessment quality of human face image |
| CN108335289A (en) * | 2018-01-18 | 2018-07-27 | 天津大学 | A kind of full image method for evaluating objective quality with reference to fusion |
| CN109376628A (en) * | 2018-10-10 | 2019-02-22 | 杭州登虹科技有限公司 | A kind of picture quality detection method, device and storage medium |
| WO2019037346A1 (en) * | 2017-08-25 | 2019-02-28 | 广州视源电子科技股份有限公司 | Method and device for optimizing human face picture quality evaluation model |
-
2019
- 2019-03-13 CN CN201910189151.9A patent/CN109977815B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060147107A1 (en) * | 2005-01-04 | 2006-07-06 | Microsoft Corporation | Method and system for learning-based quality assessment of images |
| CN102201061A (en) * | 2011-06-24 | 2011-09-28 | 常州锐驰电子科技有限公司 | Intelligent safety monitoring system and method based on multilevel filtering face recognition |
| DE102014108924A1 (en) * | 2013-06-28 | 2015-01-15 | Cognex Corporation | A semi-supervised procedure for training an auxiliary model to recognize and capture multiple patterns |
| CN103824068A (en) * | 2014-03-19 | 2014-05-28 | 上海看看智能科技有限公司 | Human face payment authentication system and method |
| CN105979253A (en) * | 2016-05-06 | 2016-09-28 | 浙江科技学院 | Generalized regression neural network based non-reference stereoscopic image quality evaluation method |
| CN107292331A (en) * | 2017-05-22 | 2017-10-24 | 浙江科技学院 | Based on unsupervised feature learning without with reference to screen image quality evaluating method |
| WO2019037346A1 (en) * | 2017-08-25 | 2019-02-28 | 广州视源电子科技股份有限公司 | Method and device for optimizing human face picture quality evaluation model |
| CN108269250A (en) * | 2017-12-27 | 2018-07-10 | 武汉烽火众智数字技术有限责任公司 | Method and apparatus based on convolutional neural networks assessment quality of human face image |
| CN108335289A (en) * | 2018-01-18 | 2018-07-27 | 天津大学 | A kind of full image method for evaluating objective quality with reference to fusion |
| CN109376628A (en) * | 2018-10-10 | 2019-02-22 | 杭州登虹科技有限公司 | A kind of picture quality detection method, device and storage medium |
Non-Patent Citations (3)
| Title |
|---|
| SEBASTIAN BOSSE ET AL: "Deep Neural Networks for No-Reference and Full-Reference Image Quality Assessment", 《 IEEE TRANSACTIONS ON IMAGE PROCESSING 》, 10 October 2017 (2017-10-10), pages 206 - 219 * |
| 李翔: "基于SVM和PSO的图像质量评价", 《计算机工程》, vol. 38, no. 23, pages 215 - 218 * |
| 汪丽丽 等: "基于Spark 的肿瘤基因混合特征选择方法", 《计算机工程》, vol. 44, no. 11, pages 1 - 6 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113012089A (en) * | 2019-12-19 | 2021-06-22 | 北京金山云网络技术有限公司 | Image quality evaluation method and device |
| CN111881732A (en) * | 2020-06-17 | 2020-11-03 | 北京邮电大学 | SVM (support vector machine) -based face quality evaluation method |
| CN112597909A (en) * | 2020-12-25 | 2021-04-02 | 北京芯翌智能信息技术有限公司 | Method and equipment for evaluating quality of face picture |
| CN113553971A (en) * | 2021-07-29 | 2021-10-26 | 青岛以萨数据技术有限公司 | Method and device for extracting optimal frame of face sequence and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109977815B (en) | 2024-07-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109977815B (en) | Image quality evaluation method and device, electronic equipment and storage medium | |
| CN108710847B (en) | Scene recognition method and device and electronic equipment | |
| CN109034078B (en) | Training method of age identification model, age identification method and related equipment | |
| US20200372243A1 (en) | Image processing method and apparatus, facial recognition method and apparatus, and computer device | |
| CN108961303B (en) | Image processing method and device, electronic equipment and computer readable medium | |
| EP3848887A1 (en) | Gan network-based vehicle damage image enhancement method and apparatus | |
| US9633044B2 (en) | Apparatus and method for recognizing image, and method for generating morphable face images from original image | |
| CN112541864A (en) | Image restoration method based on multi-scale generation type confrontation network model | |
| EP3080779A1 (en) | Efficient facial landmark tracking using online shape regression method | |
| CN107633237B (en) | Image background segmentation method, device, equipment and medium | |
| CN108269254A (en) | Image quality measure method and apparatus | |
| CN111881926A (en) | Image generation method, image generation model training method, image generation device, image generation equipment and image generation medium | |
| CN110598638A (en) | Model training method, face gender prediction method, device and storage medium | |
| CN107590460A (en) | Face classification method, apparatus and intelligent terminal | |
| CN111353364A (en) | Dynamic face identification method and device and electronic equipment | |
| JP2023526899A (en) | Methods, devices, media and program products for generating image inpainting models | |
| WO2020179200A1 (en) | Information processing method and information processing system | |
| WO2015176502A1 (en) | Image feature estimation method and device | |
| CN111814534A (en) | Visual task processing method and device and electronic system | |
| CN115205736A (en) | Video data identification method and device, electronic equipment and storage medium | |
| WO2021047453A1 (en) | Image quality determination method, apparatus and device | |
| CN113222841A (en) | Image processing method, device, equipment and medium | |
| CN115115552B (en) | Image correction model training method, image correction device and computer equipment | |
| CN110766631A (en) | Face image modification method and device, electronic equipment and computer readable medium | |
| CN114863224B (en) | Training method, image quality detection method, device and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |