CN111311581A - Image scoring method based on illumination and system and device thereof - Google Patents
Image scoring method based on illumination and system and device thereof Download PDFInfo
- Publication number
- CN111311581A CN111311581A CN202010103987.5A CN202010103987A CN111311581A CN 111311581 A CN111311581 A CN 111311581A CN 202010103987 A CN202010103987 A CN 202010103987A CN 111311581 A CN111311581 A CN 111311581A
- Authority
- CN
- China
- Prior art keywords
- picture
- scoring
- illumination
- model
- score
- 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
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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
-
- 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
- G06T2207/30196—Human being; Person
- G06T2207/30201—Face
Abstract
The invention relates to the technical field of image processing, in particular to a method, a system and a device for grading an image based on illumination. An illumination-based picture scoring method comprises the following steps: respectively inputting the picture to be scored and the target object extracted from the picture to be scored into a picture scoring model and a target object scoring model to obtain a first score and a second score; the first score and the second score are averaged to obtain an illumination value of the picture; the image scoring model and the target object scoring model are obtained by respectively grading and scoring the images under different illumination conditions and the target object extracted from the images according to illumination standards. According to the method, the relevant features of the picture can be well extracted by utilizing the convolutional neural network, and a model capable of automatically scoring is built, so that the lighting aspect of the picture is effectively scored.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a method, a system and a device for grading an image based on illumination.
Background
The scoring of pictures has wide application in real life. Whether the camera shooting is performed in the bank or the photo examination is performed in the company, the quality of the picture needs to be strictly controlled, and at this time, an automatic scoring system of the picture becomes necessary.
The current control of picture quality is either based on manual work or simply scoring some picture's basic elements (such as contrast, brightness) using conventional algorithms. The implementation of these conventional algorithms also depends heavily on the manually set threshold.
The score of the picture illumination is rich in a plurality of elements and can not be summarized simply by one light and one dark. The existing picture processing method is aimed at a single factor and lacks of integral control.
Automatic scoring designed according to a fixed standard and a fair and fair algorithm are necessary, and an efficient and effective model is not available for automatic scoring.
Disclosure of Invention
The invention aims to provide a picture grading method based on illumination, a system and a device thereof, which can well extract relevant characteristics of pictures by utilizing a convolutional neural network and build a model capable of automatically grading, thereby effectively grading the illumination aspect of the pictures.
In order to achieve the above object, a technical solution of a first aspect of the present invention provides an illumination-based picture scoring method, including the following steps:
respectively inputting the picture and the target object extracted from the picture into a picture scoring model and a target object scoring model to obtain a first score and a second score;
the first score and the second score are averaged to obtain an illumination value of the picture;
the image scoring model and the target object scoring model are obtained by respectively grading and scoring the images under different illumination conditions and the target object extracted from the images according to illumination standards.
The image scoring method based on illumination provided by the invention is characterized in that the images with different illumination conditions and the target objects extracted from the images are respectively graded and scored according to the illumination standard based on the convolutional neural network, so that an image scoring model and a target object scoring model are constructed, the illumination values of the images to be scored are respectively scored and averaged through the models, and the images can be effectively scored in the aspect of illumination.
In some possible embodiments, after scoring the picture and the target object extracted from the picture, residual error processing is performed on a convolutional layer of the convolutional neural network respectively, and a picture scoring model and a target object scoring model are constructed.
In some possible embodiments, the data on the data set is further fine-tuned through a pretest after the construction of the image scoring model and the object scoring model.
In some possible embodiments, the lighting criteria is graded to be 4 or more.
In some possible embodiments, the picture is a human image, and the target object extracted from the picture is a human head.
The technical scheme of the second aspect of the invention provides an illumination-based picture scoring system, which comprises:
the model module is used for grading and grading the pictures with different illumination conditions and the target objects extracted from the pictures according to illumination standards to construct a picture grading model and a target object grading model;
the scoring module is used for respectively inputting the picture and the target object extracted from the picture into the picture scoring model and the target object scoring model to obtain a first score and a second score;
and the illumination value module is used for calculating the mean value of the first score and the second score to obtain the illumination value of the picture.
In some possible embodiments, the model module further includes a convolutional neural network unit, and the convolutional neural network unit is configured to perform residual error processing on convolutional layers of the convolutional neural network after scoring the picture and the target object extracted from the picture.
In some possible embodiments, the system further includes an adjusting module, configured to perform fine adjustment on the data set through a pretest after the picture scoring model and the target scoring model are both constructed.
An aspect of the third aspect of the present invention provides a storage medium for storing executable instructions, where the executable instructions, when executed, implement the steps of the above-mentioned illumination-based picture scoring method.
The technical scheme of the fourth aspect of the invention provides an illumination-based picture scoring device, which comprises the storage medium.
Compared with the prior art, the invention at least has the following beneficial effects:
1. the method utilizes the convolutional neural network, sets the illumination standard gear value by integrating the illumination condition of the picture, constructs the scoring model based on the convolutional neural network, and provides good technical support for identifying the illumination condition of the picture.
2. According to the method, the image scoring model and the target object scoring model are utilized, the image to be scored is divided into two parts to be scored, the illumination value is obtained through the scoring mean value, and the illumination aspect of the image can be effectively scored.
3. The face illumination scoring system based on deep learning provided by the invention can automatically score all faces.
Drawings
The above and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:
fig. 1 shows a block flow diagram of a method for grading a picture based on illumination according to an embodiment of the present invention;
FIG. 2 is a block diagram illustrating a light based picture scoring system in accordance with an embodiment of the present invention;
fig. 3 is a block diagram illustrating another illumination-based picture scoring system according to an embodiment of the present invention.
Detailed Description
In order that the above objects, features and advantages of the present invention can be more clearly understood, a more particular description of the invention will be rendered by reference to the appended drawings. It should be noted that the embodiments of the present invention and features of the embodiments may be combined with each other without conflict.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced in other ways than those specifically described herein, and therefore the scope of the present invention is not limited to the specific embodiments disclosed below.
As shown in fig. 1, an embodiment of the present invention discloses an illumination-based picture scoring method, including the following steps:
respectively inputting the picture and the target object extracted from the picture into a picture scoring model and a target object scoring model to obtain a first score and a second score;
the first score and the second score are averaged to obtain an illumination value of the picture;
the image scoring model and the target object scoring model are obtained by respectively grading and scoring the images under different illumination conditions and the target object extracted from the images according to illumination standards.
The method comprises the steps of setting an illumination standard grade value by integrating illumination conditions of a picture based on a convolutional neural network, then respectively constructing scoring models for the picture and a target object in the picture, and respectively scoring the picture and the target object in the picture in the two models to obtain a scoring average value, wherein the scoring average value is the illumination value of the picture. Through the steps, the accuracy of identifying the illumination of the picture is effectively improved.
In some possible embodiments, after scoring the picture and the target object extracted from the picture, residual error processing is performed on a convolutional layer of the convolutional neural network respectively, and a picture scoring model and a target object scoring model are constructed.
According to the invention, a plurality of illumination standard gear values are set for the pictures shot at different time intervals and under different illumination conditions, for example, the illumination values of the pictures under different illumination shown in fig. 1 are 0.083, 0.175, 0.341, 0.667, 0.883 and 1.0 from a to b in sequence. Of course, the illumination value of the picture can also be set to other graduated values, such as 10, 20, 30, 40, 50, 60.
The pictures in the model of the invention can continuously collect new data in different scenes, including but not limited to office, cafe, market, library, family and outdoor scenes; obtaining pictures of different places and different illumination conditions; respectively extracting a target object, such as a human face, from the picture, and retaining all original information of the target object, wherein the extracted human face picture is shown in fig. 2. And then, respectively grading the pictures of different places under different illumination conditions and the faces extracted from the pictures according to illumination standards to carry out strict manual scoring, carrying out residual error processing on the scored numerical values in a convolutional layer of the convolutional neural network based on the convolutional neural network, respectively calculating regression tasks by the pictures or target objects extracted from the pictures and a data set constructed by scores, and respectively constructing to obtain a picture scoring model and a target object scoring model. The whole convolutional neural network keeps a proper depth and breadth.
It should be noted that, in the present invention, the picture and the target object extracted from the picture are respectively graded according to the illumination standard, and the grading pictures of the illumination standard can be respectively established by the picture and the grading of the illumination standard of the target object extracted from the picture, or one grading picture of the illumination standard can be shared.
In addition, residual error processing is carried out on the convolution layer of the convolution neural network, the residual error processing can be carried out on a residual error module, the residual error processing can also be replaced by a traditional convolution network module with deeper depth, and only the convolution network can obtain a better calculation result by consuming higher calculation power.
In some possible embodiments, the data on the data set is further fine-tuned through a pretest after the construction of the image scoring model and the object scoring model.
The constructed image scoring model and the target object scoring model are pre-trained on a large visual database (ImageNet), and then the numerical values on the data set are finely adjusted through pre-training respectively, so that the scoring accuracy of the image scoring model and the target object scoring model is improved.
In the present invention, in order to achieve better illumination identification, the illumination standard is graded into a plurality of numbers, such as 4, 6, 8, 10, etc.
In some possible embodiments, the lighting criteria is graded to be 4 or more.
And in the scoring process, the comprehensive illumination condition of the picture is used for scoring unless the background of the picture is dark or bright. And the target object extracted from the picture is also scored according to the comprehensive illumination condition of the target object and all the original information reserved by the target object. When the image is scored at present, the image is scored according to the brightness or the contrast of the image, and the comprehensive illumination of the image is scored according to the comprehensive illumination of the image, so that the comprehensive factors of the image are considered, and the accuracy of scoring of the illumination value of the image can be effectively improved.
The pictures in the invention can be human pictures, animal pictures or plant pictures; for example, the number of the people in the people picture can be one or more; similarly, the number of animals in the animal picture may be one or more, and the number of plants in the plant picture may be one or more. Accordingly, the object in the picture is extracted, such as the figure picture, all or part of the figure, such as the head, is extracted, the animal picture, such as the animal, is extracted, and the plant picture, such as the plant, is extracted. Accordingly, the description is omitted.
In some possible embodiments, the picture is a human image, and the target object extracted from the picture is a human head. The human head can be a human face or the human head with other part angles.
Dropout is 0.05, the convolution layer uses five layers of 128, 64, 32, 16 and 8, the same picture quantity, the grading mode of the single picture and the grading mode of the picture + the face are set, the grading mode is set to be six illumination standard files as shown in figure 1, and the accuracy of the grading mode of the picture + the face is improved by 5% compared with the grading mode of the single picture.
As shown in fig. 2, based on the above, an embodiment of the present invention further provides a system for scoring a picture based on illumination, including:
the model module is used for grading and grading the pictures with different illumination conditions and the target objects extracted from the pictures according to illumination standards to construct a picture grading model and a target object grading model;
the scoring module is used for respectively inputting the picture and the target object extracted from the picture into the picture scoring model and the target object scoring model to obtain a first score and a second score;
and the illumination value module is used for calculating the mean value of the first score and the second score to obtain the illumination value of the picture.
The invention constructs a grading method of a picture by utilizing deep learning and pattern recognition, the method is based on a convolutional neural network, sets an illumination standard gear value by integrating illumination conditions of the picture, and then constructs a grading model, namely a model module, for the picture and a target object in the picture respectively; respectively scoring the picture and the target object in the picture in the two models, namely a scoring module; and the obtained scoring and averaging value is the illumination value of the picture, namely an illumination value module. The picture is processed by the units, so that the accuracy of identifying the illumination of the picture is effectively improved.
In some possible embodiments, the model module further includes a convolutional neural network unit, and the convolutional neural network unit is configured to perform residual error processing on convolutional layers of the convolutional neural network after scoring the picture and the target object extracted from the picture.
As shown in fig. 3, in some possible embodiments, the system further includes an adjusting module, configured to perform fine adjustment on the data set through a pretest after the picture scoring model and the target scoring model are both constructed.
Based on the above-mentioned lighting-based picture scoring method, an embodiment of the present invention further provides a storage medium for storing executable instructions, which when executed implement the steps of the above-mentioned lighting-based picture scoring method.
Based on this understanding, the technical solutions of the present invention may be embodied in the form of a software product, which may be stored on an electronic device executing the methods of the various implementation scenarios of the present invention. Other modules may also be included in the storage medium.
The embodiment of the invention also provides a device for grading the pictures based on illumination, which comprises the storage medium.
The method, the system and the device for grading the pictures based on illumination can be used for recording the picture grading of a photo system, and in order to keep better identification of the portrait, the photos higher than a certain score can be set to allow recording; in addition, the system can also be used for the evaluation of photo contests, and in order to keep the better display effect of the whole picture, the evaluation given by the system can be referred to as whether the contest works are qualified or not; and so on.
In addition, it should be noted that, in the different embodiments of the present invention, the technical features in some possible implementations may be arbitrarily combined to form different embodiments. And will not be described herein.
In the present invention, the terms "first", "second", and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance; the term "plurality" means two or more unless expressly limited otherwise. The terms "mounted," "connected," "fixed," and the like are to be construed broadly, and for example, "connected" may be a fixed connection, a removable connection, an integral connection, or a virtual connection; "coupled" may be direct or indirect through an intermediary. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
The flowchart and block diagrams in the figures of the present invention illustrate the architecture, functionality, and operation of possible implementations of systems, methods and apparatus according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by systems that perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
In the description of the present specification, the description of the terms "some possible implementations" or the like means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. The method for grading the pictures based on illumination is characterized by comprising the following steps of:
respectively inputting the picture and the target object extracted from the picture into a picture scoring model and a target object scoring model to obtain a first score and a second score;
the first score and the second score are averaged to obtain an illumination value of the picture;
the image scoring model and the target object scoring model are obtained by respectively grading and scoring the images under different illumination conditions and the target object extracted from the images according to illumination standards.
2. The illumination-based picture scoring method according to claim 1, wherein the picture and the target extracted from the picture are scored and then subjected to residual error processing on convolution layers of a convolutional neural network respectively, so that a picture scoring model and a target scoring model are constructed.
3. The illumination-based picture scoring method according to claim 1, wherein the picture scoring model and the target scoring model are constructed and then used for fine-tuning data on the data set through a pretest.
4. A lighting-based picture scoring method according to any one of claims 1-3, wherein the lighting criteria are ranked by 4 or more.
5. The illumination-based picture scoring method according to any one of claims 1-3, wherein the picture is a human image and the object extracted from the picture is a human head.
6. A lighting-based picture scoring system, comprising:
the model module is used for grading and grading the pictures with different illumination conditions and the target objects extracted from the pictures according to illumination standards to construct a picture grading model and a target object grading model;
the scoring module is used for respectively inputting the picture and the target object extracted from the picture into the picture scoring model and the target object scoring model to obtain a first score and a second score;
and the illumination value module is used for calculating the mean value of the first score and the second score to obtain the illumination value of the picture.
7. The illumination-based picture scoring system according to claim 6, wherein the model module further comprises a convolutional neural network unit for performing residual error processing on convolutional layers of a convolutional neural network after scoring the picture and the target extracted from the picture, respectively.
8. The illumination-based picture scoring system according to claim 6, further comprising an adjustment module for fine-tuning the dataset by pre-testing after the picture scoring model and the object scoring model are both constructed.
9. A storage medium storing executable instructions that when executed perform the steps of the illumination-based picture scoring method of any one of claims 1-5.
10. An illumination-based picture scoring apparatus comprising the storage medium of claim 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010103987.5A CN111311581A (en) | 2020-02-20 | 2020-02-20 | Image scoring method based on illumination and system and device thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010103987.5A CN111311581A (en) | 2020-02-20 | 2020-02-20 | Image scoring method based on illumination and system and device thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111311581A true CN111311581A (en) | 2020-06-19 |
Family
ID=71161854
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010103987.5A Pending CN111311581A (en) | 2020-02-20 | 2020-02-20 | Image scoring method based on illumination and system and device thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111311581A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018049952A1 (en) * | 2016-09-14 | 2018-03-22 | 厦门幻世网络科技有限公司 | Photo acquisition method and device |
| CN108805970A (en) * | 2018-05-03 | 2018-11-13 | 百度在线网络技术（北京）有限公司 | illumination estimation method and device |
| 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 |
| CN109635664A (en) * | 2018-11-15 | 2019-04-16 | 珠海研果科技有限公司 | A kind of method for detecting fatigue driving based on illumination detection |
| CN109919073A (en) * | 2019-03-01 | 2019-06-21 | 中山大学 | A kind of recognition methods again of the pedestrian with illumination robustness |
-
2020
- 2020-02-20 CN CN202010103987.5A patent/CN111311581A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018049952A1 (en) * | 2016-09-14 | 2018-03-22 | 厦门幻世网络科技有限公司 | Photo acquisition method and device |
| 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 |
| CN108805970A (en) * | 2018-05-03 | 2018-11-13 | 百度在线网络技术（北京）有限公司 | illumination estimation method and device |
| CN109635664A (en) * | 2018-11-15 | 2019-04-16 | 珠海研果科技有限公司 | A kind of method for detecting fatigue driving based on illumination detection |
| CN109919073A (en) * | 2019-03-01 | 2019-06-21 | 中山大学 | A kind of recognition methods again of the pedestrian with illumination robustness |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109714519B (en) | Method and system for automatically adjusting image frame | |
| CN102576461B (en) | The aesthetic quality of assessment digital picture | |
| CN110210276A (en) | A kind of motion track acquisition methods and its equipment, storage medium, terminal | |
| CN108288027A (en) | A kind of detection method of picture quality, device and equipment | |
| CN110807757B (en) | Image quality evaluation method and device based on artificial intelligence and computer equipment | |
| CN108564052A (en) | Multi-cam dynamic human face recognition system based on MTCNN and method | |
| CN110807759B (en) | Method and device for evaluating photo quality, electronic equipment and readable storage medium | |
| CN109615627B (en) | Power transmission and transformation inspection image quality evaluation method and system | |
| CN112287827A (en) | Complex environment pedestrian mask wearing detection method and system based on intelligent lamp pole | |
| CN111833322B (en) | Garbage multi-target detection method based on improved YOLOv3 | |
| CN107103608B (en) | A kind of conspicuousness detection method based on region candidate samples selection | |
| CN109120919A (en) | A kind of automatic analysis system and method for the evaluation and test of picture quality subjectivity | |
| CN102567734A (en) | Specific value based retina thin blood vessel segmentation method | |
| CN110782448A (en) | Rendered image evaluation method and device | |
| CN115689928B (en) | Method and system for removing duplication of transmission tower inspection images under visible light | |
| CN112449115B (en) | Shooting method and device and electronic equipment | |
| CN111667511B (en) | Method, device and system for extracting background in dynamic video | |
| CN112906510A (en) | Fishery resource statistical method and system | |
| CN111311581A (en) | Image scoring method based on illumination and system and device thereof | |
| WO2023217138A1 (en) | Parameter configuration method and apparatus, device, storage medium and product | |
| CN116681613A (en) | Illumination-imitating enhancement method, device, medium and equipment for face key point detection | |
| CN111031258A (en) | Lunar vehicle navigation camera exposure parameter determination method and device | |
| CN110796650A (en) | Image quality evaluation method and device, electronic equipment and storage medium | |
| CN115909306A (en) | Method and device for determining polishing scheme | |
| CN115147392A (en) | Improved YOLOv 4-based coal gangue detection method and 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 |