CN111127456A - Image annotation quality evaluation method - Google Patents
Image annotation quality evaluation method Download PDFInfo
- Publication number
- CN111127456A CN111127456A CN201911383370.7A CN201911383370A CN111127456A CN 111127456 A CN111127456 A CN 111127456A CN 201911383370 A CN201911383370 A CN 201911383370A CN 111127456 A CN111127456 A CN 111127456A
- Authority
- CN
- China
- Prior art keywords
- image
- quality
- map
- data set
- evaluation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 18
- 238000013441 quality evaluation Methods 0.000 title description 6
- 238000011156 evaluation Methods 0.000 claims abstract description 25
- 238000004364 calculation method Methods 0.000 claims abstract description 15
- 238000012706 support-vector machine Methods 0.000 claims abstract description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 238000012935 Averaging Methods 0.000 claims description 2
- 230000000007 visual effect Effects 0.000 abstract description 4
- 238000011002 quantification Methods 0.000 abstract description 2
- 238000002372 labelling Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000001303 quality assessment method Methods 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000016776 visual perception Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method for evaluating image annotation quality, which comprises the following steps: s1, extracting MSCN coefficient from the image to be evaluated; s2, fitting the MSCN coefficient into asymmetric generalized Gaussian distribution, and extracting the characteristics of the fitted asymmetric generalized Gaussian distribution; s3, inputting the extracted features into a support vector machine for regression to obtain a distortion data set of the image; s4, MAP calculation is carried out on the distortion data set, the calculation results are averaged to obtain the evaluation result of the image quality, the method uses a computer to replace a human visual system to watch and recognize images, and the quality of the consistency of the image quality quantification value and the human subjective observation value is obtained through a calculation model to evaluate.
Description
Technical Field
The present invention relates to the field of image evaluation. And more particularly, to a method for image annotation quality assessment.
Background
The image annotation refers to a process of automatically adding text characteristic information reflecting the content of an image to the image by a machine learning method aiming at the visual content of the image. The basic idea is to automatically learn the potential association or mapping relation between the semantic concept space and the visual feature space by using the labeled image set or other available information, and add text keywords to the unknown image. Through the processing of the automatic image annotation technology, the image information problem can be converted into the text information processing problem with relatively mature technology.
Image labeling based on classification is a supervised machine learning approach. The classifier training process can continuously adjust the classifier through feedback information, so that the classifier achieves certain precision. The basic idea of the classification model is to segment the image, filter noise and over-segmentation parts, regard each semantic concept as a class, and classify the segmented image. Automatic labeling of images can actually be handled as an image classification problem.
The automatic labeling method of the related model image is based on an early probability correlation model, and is different from the probability correlation model in that the method not only simply counts the symbiotic probability of an image area and the occurrence of a keyword, but also establishes a probability correlation model between the image and a semantic keyword. And finding a group of semantic keywords with the maximum relevance probability for the image to be labeled through the association model to label the image.
Disclosure of Invention
One object of the present invention is to provide a method for evaluating image annotation quality, comprising the following steps:
s1, extracting MSCN coefficient from the image to be evaluated;
s2, fitting the MSCN coefficient into asymmetric generalized Gaussian distribution, and extracting the characteristics of the fitted asymmetric generalized Gaussian distribution;
s3, inputting the extracted features into a support vector machine for regression to obtain a distortion data set of the image;
and S4, performing MAP calculation on the distortion data set, and averaging the calculation results to obtain the evaluation result of the image quality.
Preferably, the step S4 further includes, before the MAP calculation, putting a weight factor K in a MAP algorithm for calculating an accurate ranking value of the distortion data set.
Preferably, the step of placing the weight factor K in the MAP algorithm comprises:
weighting factor knInitially set to 1/N, wherein
ki+1＝argkminKL(k|ki)+μhi(k)
argkmin is a minimum function, KL is a KL difference, hi(k) Is a hinge loss function;
h isi(k) Obtained by the following formula
Si＝(S1(x，Г+)-S1(x，Г-)，…，SN(x，Г+)-SM(x，Г-))
Wherein, Г+And Γ-Is the image content;
subjecting the said SiAnd (3) placing the data samples into the MAP, classifying each data sample, and recording the following conditions:
TP, FN, FP, TN, calculate an Accuracy Accuracy:
and multiplying the Accuracy by the weighting factor k to finish the evaluation of the image annotation quality.
The invention has the following beneficial effects:
the invention provides a method for evaluating image labeling quality, which is an image relative evaluation method fusing evaluation factors on the basis of BRISQUE and MAP algorithms, wherein MSCN is raised from an image, coefficients of the MSCN are fit to AGGD asymmetric generalized Gaussian distribution, the characteristics of the fit Gaussian distribution are extracted and input to a Support Vector Machine (SVM) for regression, then MAP calculation is carried out, namely, the reciprocal of the sequence of the result given in an evaluated system is taken as the accuracy of the result, and then the average value is taken, so that the evaluation result of the image quality is obtained.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 shows a flow diagram of a method of annotation quality assessment of the present invention.
Detailed Description
In order to make the technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
It should be noted that the term [ the embodiment of the present invention ] does not limit the execution sequence of steps a to e, and for example, step c and step a ] can be executed in sequence
All numerical designations of the invention (e.g., temperature, time, concentration, weight, and the like, including ranges for each) may generally be approximations that vary (+) or (-) in increments of 0.1 or 1.0, as appropriate. All numerical designations should be understood as preceded by the term "about". "C (B)
The invention provides an image relative evaluation method based on BRISQUE and MAP algorithm and integrating evaluation factors.
In the case where the image information technology is widely used, evaluation of image quality becomes a wide and fundamental problem. Since image information has incomparable advantages over other information, rational processing of image information is an indispensable means in various fields. In the process of acquiring, processing, transmitting and recording images, due to the imperfection of an imaging system, a processing method, a transmission medium, recording equipment and the like, and the reasons of object motion, noise pollution and the like, certain image distortion and degradation are inevitably brought, which brings great difficulty for people to know an objective world and research and solve problems.
For example, in image annotation, the accuracy and reliability of an identification result are directly affected by the accurate annotation of an acquired image; for another example, the systems such as the teleconference and the video on demand are affected by adverse factors such as transmission errors and network delay, and online real-time image quality monitoring is needed, so that a service provider can dynamically adjust the information source positioning strategy, and the requirement of service quality is further met; in military applications, the effectiveness of battlefield surveillance and combat assessment is also dependent on the quality of the images or videos captured by aerial equipment such as drones. Therefore, the reasonable evaluation on the image annotation quality has very important application value.
The image labeling quality evaluation method is distinguished from the aspect of existence of people, and comprises two branches of subjective evaluation and objective evaluation. The subjective evaluation takes a person as an observer, carries out subjective evaluation on the image and strives to truly reflect the visual perception of the person; the objective evaluation method reflects subjective perception of human eyes by means of a certain mathematical model and provides a result based on digital calculation.
The subjective evaluation only relates to the qualitative evaluation made by people, and the subjective qualitative evaluation is made on the quality of the image by people. The selection for the observer typically takes into account either an untrained "outer row" or a trained "inner row". The method is established in the statistical sense, and enough observers should participate in evaluation to ensure that the subjective evaluation of the image is statistically significant.
The basic goal of objective evaluation of image quality is to design a computational model that accurately and automatically perceives image quality. The ultimate goal is the desire to replace the human visual system with a computer to view and perceive images. Internationally, objective evaluation of image quality is usually evaluated by testing the performance of a plurality of factors affecting image quality and obtaining the consistency between an image quality quantification value and a human subjective observation value through a calculation model. Imatest in the united states and DxO analyzer in france are objective image quality evaluation systems to which the names are compared.
Examples
Example 1
BRISQE is a spatial domain image quality evaluation algorithm without reference;
an image labeling quality evaluation method as shown in fig. 1 is to project MSCN from an image, fit coefficients thereof to AGGD asymmetric generalized gaussian distribution, extract features of the fitted gaussian distribution, input the features to a support vector machine SVM for regression, then perform MAP calculation, that is, take the reciprocal of the ranking of the results given in the evaluated system as the accuracy thereof, and then take an average value, thereby evaluating the image quality.
The innovation point of the invention is that the introduced weight factor k is setCalculating the accurate ranking value in MAP calculation, wherein k isnInitially set to 1/N, after which the following is performed:
ki+1＝argkminKL(k|ki)+μhi(k)
wherein KL represents a KL difference, hi(k) Representing the hinge loss function:
wherein S isiComprises the following steps:
Si＝(S1(x，Г+)-S1(x，Г-)，…，SN(x，Г+)-SM(x，Г-))
therein, label Г+And Γ-Reflecting the content of the image.
Calculating S from the result of the calculationiAnd (3) placing the data samples into the MAP, classifying each data sample, and recording several conditions: TP, FN, FP, TN, calculate an Accuracy Accuracy:
the image annotation quality can be evaluated by multiplying Accuracy by a weighting factor k.
It should be understood that the above-mentioned embodiments of the present invention are only examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention, and it will be obvious to those skilled in the art that other variations or modifications may be made on the basis of the above description, and all embodiments may not be exhaustive, and all obvious variations or modifications may be included within the scope of the present invention.
Claims (3)
1. A method for evaluating image annotation quality is characterized by comprising the following steps:
s1, extracting MSCN coefficient from the image to be evaluated;
s2, fitting the MSCN coefficient into asymmetric generalized Gaussian distribution, and extracting the characteristics of the fitted asymmetric generalized Gaussian distribution;
s3, inputting the extracted features into a support vector machine for regression to obtain a distortion data set of the image;
and S4, performing MAP calculation on the distortion data set, and averaging the calculation results to obtain the evaluation result of the image quality.
2. The method according to claim 1, wherein the step S4 further comprises placing a weighting factor K in a MAP algorithm for calculating an accurate ranking value of the distortion data set before the MAP calculation.
3. The method of claim 2, wherein the placing of the weight factor K in the MAP algorithm comprises:
weighting factor knInitially set to 1/N, wherein
ki+1＝argkminKL(k|ki)+μhi(k)
argkmin is a minimum function, KL is a KL difference, hi(k) Is a hinge loss function;
h isi(k) Obtained by the following formula
Si＝(S1(x,Γ+)-S1(x,Γ-),…,SN(x,Γ+)-SM(x,Γ-))
Wherein, gamma is+And Γ-Is the image content;
subjecting the said SiAnd (3) placing the data samples into the MAP, classifying each data sample, and recording the following conditions:
TP, FN, FP, TN, calculate an Accuracy Accuracy:
and multiplying the Accuracy by the weighting factor k to finish the evaluation of the image annotation quality.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911383370.7A CN111127456A (en) | 2019-12-28 | 2019-12-28 | Image annotation quality evaluation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911383370.7A CN111127456A (en) | 2019-12-28 | 2019-12-28 | Image annotation quality evaluation method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111127456A true CN111127456A (en) | 2020-05-08 |
Family
ID=70505319
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911383370.7A Pending CN111127456A (en) | 2019-12-28 | 2019-12-28 | Image annotation quality evaluation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111127456A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111753873A (en) * | 2020-05-12 | 2020-10-09 | 北京捷通华声科技股份有限公司 | Image detection method and device |
| CN111915559A (en) * | 2020-06-30 | 2020-11-10 | 电子科技大学 | Airborne SAR image quality evaluation method based on SVM classification credibility |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105184303A (en) * | 2015-04-23 | 2015-12-23 | 南京邮电大学 | Image marking method based on multi-mode deep learning |
| EP2977933A1 (en) * | 2014-07-22 | 2016-01-27 | Baden-Württemberg Stiftung gGmbH | Image classification |
| CN106815839A (en) * | 2017-01-18 | 2017-06-09 | 中国科学院上海高等研究院 | A kind of image quality blind evaluation method |
-
2019
- 2019-12-28 CN CN201911383370.7A patent/CN111127456A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2977933A1 (en) * | 2014-07-22 | 2016-01-27 | Baden-Württemberg Stiftung gGmbH | Image classification |
| CN105184303A (en) * | 2015-04-23 | 2015-12-23 | 南京邮电大学 | Image marking method based on multi-mode deep learning |
| CN106815839A (en) * | 2017-01-18 | 2017-06-09 | 中国科学院上海高等研究院 | A kind of image quality blind evaluation method |
Non-Patent Citations (1)
| Title |
|---|
| LI SONG 等: "《Blind Image Quality Assessment Based On a New Feature of Nature Scene Statistics》" * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111753873A (en) * | 2020-05-12 | 2020-10-09 | 北京捷通华声科技股份有限公司 | Image detection method and device |
| CN111915559A (en) * | 2020-06-30 | 2020-11-10 | 电子科技大学 | Airborne SAR image quality evaluation method based on SVM classification credibility |
| CN111915559B (en) * | 2020-06-30 | 2022-09-20 | 电子科技大学 | Airborne SAR image quality evaluation method based on SVM classification credibility |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2020010785A1 (en) | Classroom teaching cognitive load measuring system | |
| CN110532970B (en) | Age and gender attribute analysis method, system, equipment and medium for 2D images of human faces | |
| US20190026884A1 (en) | Method for assessing aesthetic quality of natural image based on multi-task deep learning | |
| CN111344713A (en) | Camera and image calibration for object recognition | |
| CN112183334B (en) | Video depth relation analysis method based on multi-mode feature fusion | |
| CN104899195B (en) | Recommendation method and device for personalized education resources | |
| CN111127456A (en) | Image annotation quality evaluation method | |
| CN112348809B (en) | No-reference screen content image quality evaluation method based on multitask deep learning | |
| CN112711676B (en) | Video recall method and device, electronic equipment and storage medium | |
| CN110688471A (en) | Training sample obtaining method, device and equipment | |
| CN117809124B (en) | Medical image association calling method and system based on multi-feature fusion | |
| CN108257117B (en) | Image exposure evaluation method and device | |
| CN107169093B (en) | target image acquisition method and device | |
| CN112949456A (en) | Video feature extraction model training method and device, and video feature extraction method and device | |
| CN111462087B (en) | Image detection method, device and system based on artificial intelligence and storage medium | |
| CN113298161A (en) | Image recognition model testing method and device, computer equipment and storage medium | |
| CN110598607B (en) | Non-contact and contact cooperative real-time emotion intelligent monitoring system | |
| CN112633037A (en) | Object monitoring method and device, storage medium and electronic equipment | |
| Yan et al. | No-reference remote sensing image quality assessment based on gradient-weighted natural scene statistics in spatial domain | |
| CN111612021A (en) | Error sample identification method and device and terminal | |
| CN111259698A (en) | Method and device for acquiring image | |
| CN115116117A (en) | Learning input data acquisition method based on multi-mode fusion network | |
| CN117218570A (en) | Multi-person attendance checking method, device, attendance checking equipment and storage medium | |
| CN110059584B (en) | Event naming method combining boundary distribution and correction | |
| CN113591767A (en) | Method and device for determining image recognition evaluation value, storage medium and electronic device |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200508 |