CN113393388A - Image enhancement method, device adopting same, storage medium and vehicle - Google Patents
Image enhancement method, device adopting same, storage medium and vehicle Download PDFInfo
- Publication number
- CN113393388A CN113393388A CN202110578233.XA CN202110578233A CN113393388A CN 113393388 A CN113393388 A CN 113393388A CN 202110578233 A CN202110578233 A CN 202110578233A CN 113393388 A CN113393388 A CN 113393388A
- Authority
- CN
- China
- Prior art keywords
- data
- data set
- image
- enhancement
- sample library
- 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 70
- 238000003860 storage Methods 0.000 title claims abstract description 12
- 238000001514 detection method Methods 0.000 claims abstract description 10
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 8
- 230000007547 defect Effects 0.000 claims description 62
- 238000012545 processing Methods 0.000 claims description 23
- 238000000605 extraction Methods 0.000 claims description 18
- 230000002950 deficient Effects 0.000 claims description 12
- 238000013507 mapping Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 8
- 238000013075 data extraction Methods 0.000 claims description 7
- 230000009466 transformation Effects 0.000 claims description 6
- 239000012535 impurity Substances 0.000 claims description 4
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 3
- 239000002184 metal Substances 0.000 claims description 3
- 239000000126 substance Substances 0.000 claims 1
- 230000009286 beneficial effect Effects 0.000 abstract 1
- 239000000463 material Substances 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 4
- 238000009826 distribution Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000009776 industrial production Methods 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000006467 substitution reaction Methods 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
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
Abstract
The application discloses an image enhancement method, a device adopting the method, a computer-readable storage medium and a vehicle. The method is used for effectively expanding the detection data with smaller sample space, and is beneficial to the realization of the intelligent algorithm of related workers. Specifically, a first sample library is generated by acquiring data which accords with preset characteristics in a first data set; performing random enhancement on the first sample library to generate a second sample library; extracting data in the first data set or the second data set to generate a third data set; and combining the data in the third data set with the randomly extracted data in the second sample library according to a preset mode to generate a fourth data set. In addition, the application also discloses a device, a device and a vehicle adopting the method, and the device, the device and the vehicle can be quickly transplanted and applied to relevant occasions.
Description
Technical Field
The invention belongs to the technical field of image processing, and particularly relates to an image enhancement method, a device adopting the method, a computer-readable storage medium and a vehicle.
Background
In industrial production, the defect rate of products is often controlled to a very low level; therefore, the defect samples directly obtained from the defective products are limited. In order to meet the requirements of deep learning neural network model training and application, the problem needs to be solved urgently.
Disclosure of Invention
The application discloses an image enhancement method, and an intelligent device, a storage medium and a vehicle adopting the method. The method comprises the steps of generating a first sample library by acquiring data which are in accordance with preset characteristics in a first data set; the sample library is used for storing data with preset characteristics, and the image enhancement is realized through the cooperation of other steps.
In order to simulate the image feature distribution which may randomly appear in an actual system, the first sample library is randomly enhanced, so that a second sample library is generated; the enhanced second sample pool achieves a larger sample space than the first sample pool.
Further processing the data to be processed: generating a third data set by extracting data in the first data set or the second data set; and then combining the data in the third data set with the randomly extracted data in the second sample library according to a preset mode to generate a fourth data set subjected to enhancement processing. During the processes of acquisition, extraction, random enhancement, merging, etc., the necessary intermediate data and final data are saved for further processing.
Furthermore, the application discloses a corresponding image data enhancement method for the image in the field of industrial part defect visual detection; under the scenes, the deep learning neural network training effect of the industrial part defect data set is improved.
In an application scene of image processing, a first data set, a second data set, a third data set and a fourth data set processed by the method are all sets of image data; the first data set is a set of image data conforming to the predetermined features, the second data set is a set of image data not containing the predetermined features, the third data set is a set of image data to be enhanced, and the fourth data set is a set of images subjected to image enhancement processing.
In the visual or related scene of the industrial part defect, the first data set refers to a set of images of the part with the defect; the second data set is a set of images of the parts when the parts have no defects; the third data set refers to a set of part images of the image to be enhanced; the fourth data set refers to a set of part images after image enhancement processing. For the automotive industry scenario, the components include automotive components.
In the application scenario of the application, when a part meets a predetermined characteristic, it indicates that the part has a defect; the defects at least comprise flash, scratch, stain, deformation and impurities, but the defect state and characteristics can be redefined or set according to the process requirements of the parts; that is, these predetermined features are extensible.
Further, the random enhancement disclosed in the present application includes at least one of scaling, rotation transformation and brightness adjustment, and the algorithm adopted by the random enhancement is pre-specified or randomly selected; at this time, the algorithm adopted by the random enhancement can be redesigned according to the process requirements.
In the process of image enhancement, data which is not credible or impossible to appear in a real system is avoided, namely, after the merging processing, defects only exist in a region where the defects possibly appear; an increase in sample space at the expense of sample quality; meanwhile, the selection process of the area for embedding the enhanced image data is random; therefore, the distribution rule of the defects in the real system is further simulated. If the selection of the region is performed according to the probability distribution of the occurrence of a certain type of defect, a sample space meeting a certain constraint condition may also be obtained, so as to perform a simulation with a higher degree of conformity on data in a specific application environment.
The use of the method also considers the geometric characteristics of the part, and the method can be suitable for the part if the image observed from a certain direction of the part contains symmetrical characteristics or asymmetrical characteristics; the random enhancement means that the local area of the observed image view or the observed image is randomly rotated by an angle; when the image contains annotations, the process of random enhancement should include the process of labeling a part of the image.
Further, when the above symmetric features are circles, rings or other similar shapes and lines, the rule followed by generating the first sample library is not limited to the gray level binarization defect extraction and/or sobel operator contour extraction rule; for some parts, scratch defects may exist, and the scratch can be found by extracting the edge through a Sobel operator; for some parts, the color of a possible defect is greatly different from the non-defective color, and at the moment, the region of interest can be extracted through gray level binarization, so that the calculation amount is smaller compared with that of a Grub method, and manual initialization is not needed.
In some scenarios, the first data set is an empty set, and new data of the sample space is obtained by rotating the defect-free part.
Due to the rapid advance of the material industry, the material of the parts is not limited to plastic or metal, so long as the material can meet the requirements of industrial application, and the method can be applied as long as the acquisition of the image is feasible when relevant detection is needed. The object processed by the method is a part image, and the material difference has no substantial influence on the method; that is, different materials can be applied to the method.
Furthermore, the application discloses a defect detection device, which comprises a feature extraction unit, a random enhancement unit, a data extraction unit and a random mapping unit, wherein the feature extraction unit is used for extracting the feature of the defect; the feature extraction unit acquires data which accord with preset features in the first data set, and generates a first sample library; a random enhancement unit carries out random enhancement on the first sample library to generate a second sample library; the data extraction unit extracts data in the first data set or the second data set to generate a third data set; and the random mapping unit combines the data in the third data set and the randomly extracted data in the second sample library according to a preset mode to generate a fourth data set. After necessary data are obtained, the detection device can be applied to an artificial intelligence or similar unit, and effective detection results are output.
Still further, for the above-mentioned detection device, the microprocessor and the memory electrically connected thereto, if the memory therein is used for storing a computer program; which computer program, when being executed by a microprocessor, is adapted to carry out the respective method of the present application.
In other scenarios, the readable storage medium includes a storage medium body for storing a computer program, and the computer program is used for implementing the image enhancement method of the present application when being executed by a microprocessor. If further, when the vehicle adopts the above-mentioned device or means, it will also become an embodiment of the present application.
The method and the device provide an effective data enhancement mode aiming at the image in the field of industrial part defect visual detection, and can improve the deep learning neural network training effect of an industrial part defect data set in some scenes. The recall rate is higher (i.e., fewer missed calls) because the enhanced data set has more defective pictures.
Drawings
To more clearly explain the technical solutions of the present application and to facilitate a further understanding of the technical effects, technical features and objects of the present application, the present application will be described in detail with reference to the accompanying drawings, which form an essential part of the present specification, and which are used to explain the technical solutions of the present application together with the embodiments of the present application, but do not limit the present application.
The same reference numerals in the drawings denote the same elements, and in particular:
FIG. 1 is a schematic diagram of an image enhancement process according to an embodiment of the present application;
FIG. 2 is a block diagram of image data in a first data set and a plurality of samples in a first sample library generated after feature extraction from the image data according to example 1 of the present application;
FIG. 3 is a first sample from a first sample library according to example 1 of the present application;
FIG. 4 is a second sample from the first sample library of example 1 of the present application;
FIG. 5 shows a third sample from the first sample library of example 1 of the present application;
FIG. 6 shows a fourth sample from the first sample library of example 1 of the present application;
fig. 7 is image data in the first data set, i.e. an image before enhancement processing, according to embodiment 1 of the present application;
fig. 8 is image data in a fourth data set of embodiment 1 of the present application, namely, an image after enhancement processing;
FIG. 9 shows a first sample from a first sample library according to example 2 of the present application;
fig. 10 is an example 60 of a fourth data set in embodiment 2 of the present application, that is, an image obtained by clockwise rotating an original image by 60 degrees;
fig. 11 is an example 120 in the fourth data set of embodiment 2 of the present application, that is, an image after 120 degrees is selected clockwise from an original image;
FIG. 12 is an image of a part 201 which is regarded as non-defective according to a predetermined standard in example 3 of the present application;
FIG. 13 is an image of a part 202 of example 4 of the present application, which is considered to be non-defective according to a predetermined standard;
FIG. 14 is an image of a part 101 which is regarded as being non-defective according to a predetermined standard in example 5 of the present application;
FIG. 15 is an image of a part 102 that is considered to be non-defective according to a predetermined standard in example 6 of the present application;
wherein:
100-a feature extraction unit; 200- -random enhancement unit;
300- -a data extraction unit; 400- -random mapping unit.
Detailed Description
The present application will be described in further detail with reference to the following drawings and examples. Of course, the following description of the specific embodiments is merely for the purpose of illustrating the technical solutions of the present application and is not intended to limit the present application. In addition, the portions shown in the examples or the drawings are merely illustrative of the relevant portions of the present application, and are not all the present application.
As shown in fig. 1, it is a schematic diagram of an image enhancement process according to an embodiment of the present application; generating a first sample library by acquiring data in the first data set which accord with the preset characteristics; randomly enhancing the first sample library to generate a second sample library; extracting data in the first data set or the second data set to generate a third data set; and combining the data in the third data set with the randomly extracted data in the second sample library according to a preset mode to generate a fourth data set.
Wherein the first, second, third and fourth data sets are sets of image data; and the first data set is a collection of image data conforming to a predetermined characteristic; the second data set is a set of image data that does not contain the predetermined feature; the third data set is a set of image data to be enhanced; the fourth data set is a set of images that have undergone image enhancement processing.
As shown in fig. 2, in embodiment 1 of the present application, after image data in a first data set and features extracted from the image data, a plurality of samples in a first sample library are generated;
fig. 3-6 are the first to the fourth samples in the first sample library according to example 1 of the present application. The above samples are obtained by feature extraction from the first data set as shown in fig. 2; the extraction process may include operations such as identification, segmentation, storage, etc. of particular features.
Fig. 7 shows image data in the first data set, i.e., an image before enhancement processing in embodiment 1 of the present application; fig. 8 is image data in a fourth data set, i.e., an enhanced image, according to embodiment 1 of the present application; in contrast, a new defective data area appears in fig. 8. These regions are implemented by subjecting the data or data of fig. 3-6 to random scaling, rotation transformation, brightness adjustment, and the like.
In embodiment 1, the first data set refers to a set of images of a part with defects, and as shown in fig. 2, the defects are divided by a rectangular frame and become elements in the second sample library after subsequent storage and other necessary processing.
As another data source, the second data set refers to a set of images of the part without defects; as shown in fig. 7, the image is the image to be enhanced in the third data set in embodiment 1, and if 8, the image is an element in the fourth data set and is the image after the image enhancement processing.
FIG. 7 is a diagram showing an original image of an automobile part according to example 1 of the present application, wherein defects including, but not limited to, flash, scratch, stain, deformation, impurities, etc. may occur on the part; in order to ensure the applicability and the expandability of the method, the defect set can be expanded according to the process requirements of the parts, and an expanded feature set is obtained.
After the samples shown in fig. 3-6 are randomly enhanced, the defect feature images after the processing of scale scaling, rotation transformation, brightness adjustment and the like are combined into fig. 7, and a new sample as shown in fig. 8 is obtained, so that the expansion of the sample space is realized. The algorithm adopted in the random enhancement process is pre-designated or randomly selected; the algorithm can be redesigned according to the process requirements; meanwhile, in the enhancement process, the defect features added in the merging process can be only added in the science and believing or possible areas, so that the effectiveness of the newly added data in the sample space is ensured. Since the addition process is random, the spatial distribution of the actual product in the presence of defects is well simulated.
In embodiment 2 of the present application, processing was performed for a part having a circular feature, but the method of the present application is still applicable to an image of an asymmetrical part.
Fig. 9 is an original sample, fig. 10 and 11 are new samples after being expanded by the method of the present application, and the capacity of the sample space in embodiment 2 is expanded by the implementation of the method, where fig. 10 and 11 are obtained by rotating the original image by 60 degrees or 120 degrees along the approximate geometric center, respectively. If the original image of fig. 9 is processed by a plurality of random enhancement processes, an expanded sample space can be obtained.
In embodiment 1, a defect sample is randomly extracted, and after random size conversion and scale conversion, the converted defect image data is added to an original image; it can be seen that the enhanced local defect is very real, and through testing, the enhanced data set and the original data set can be improved by 1-9 percentage points on the recall rate of the defect:
in the embodiment 1, the extraction process of the feature data has a large difference with the surrounding background due to a general defect sample; therefore, a Grub algorithm can be used for obtaining a communication domain inside the defect precise outline; generating more possible defect states by the acquired defect samples in a mode of scaling, rotation transformation and brightness adjustment; randomly mapping the defect samples after random enhancement to a data set to be enhanced, wherein the mapping positions need to appear in areas where defects possibly appear; so far, the enhanced picture sample contains the simulated defects of the defect map.
As shown in fig. 9-11, in embodiment 2, the random enhancement process is to randomly rotate an angle for the observed image or the local area of the image; furthermore, when the observed image contains an annotation, the randomly enhanced processing should include processing of the annotation. In example 2, the symmetric feature is a circle or a torus.
Through the rotation enhancement method in the embodiment 2, defects in the sample can appear at various positions of the circular part, so that the neural network model can have stronger generalization capability. Tests have shown that the same rotation enhancement data set as the original data set can also improve the defect recall by 1-9 percentage points.
Example 2 is based on that there are many round parts in the industrial production, also include plastic or metal parts of different materials, etc.; because defects such as flash, scratch, stain, deformation, impurities and the like can appear at any position on the circular part, the method can be used for rotating the defective part to reinforce under the condition that the amount of a defective sample is small; meanwhile, if the labels exist, all labels in the image need to rotate around the axis. This can enhance the more defective work pieces obtained on the circular type part image.
The implementation 2 discloses a method for expanding generalization and enhancing to other normal or abnormal parts aiming at industrial part defects to obtain more defect samples; the method comprises the steps of acquiring a defect interesting region from a part image, and compared with the method of directly intercepting the defect interesting region, the method can better acquire the precise outline of the defect;
in addition, embodiment 2 also discloses a method for generalizing a defect region of interest sample, which combines size scaling, scaling and rotation transformation; any one, two or three of the combined enhancement modes can be selected; for a scene for sample enhancement of the circular industrial part, more defect samples can be obtained, so that the position of the defect can appear at any angle of the circular part.
Fig. 1 is a schematic diagram of an image enhancement process according to an embodiment of the present application, where the process implements image enhancement processing in cooperation with a feature extraction unit 100, a random enhancement unit 200, a data extraction unit 300, and a random mapping unit 400; wherein; the feature extraction unit 100 acquires data in a first data set, which conforms to a predetermined feature, and generates a first sample library; the random enhancement unit 200 performs random enhancement on the first sample library to generate a second sample library; the data extraction unit 300 extracts data in the first data set or the second data set to generate a third data set; the random mapping unit 400 combines the data in the third data set with the randomly drawn data in the second sample library in a predetermined manner to generate a fourth data set.
The above embodiment 1 and embodiment 2 describe the present application in detail in an application scenario of vehicle part defect detection, and as can be seen from actual measurement data and simulation results, the method effectively expands the sample capacity of data, and can improve the training effect of such part defect data sets in deep learning neural network application. .
Wherein, for the defects of flash, deformation, foreign matters and the like, the overall appearance of the product has larger change than the background, and the product can be extracted by a foreground extraction method such as Grub; circular components, rotational enhancements may be used.
It should be noted that the foregoing examples are only for clearly illustrating the technical solutions of the present application, and those skilled in the art will understand that the embodiments of the present application are not limited to the above, and obvious changes, substitutions or replacements can be made based on the above contents without departing from the scope covered by the technical solutions of the present application; other embodiments will fall within the scope of the present application without departing from the inventive concepts of the present application.
Claims (15)
1. An image enhancement method, comprising:
acquiring data which accords with preset characteristics in a first data set, and generating a first sample library;
randomly enhancing the first sample library to generate a second sample library;
extracting data in the first data set or the second data set to generate a third data set;
and combining the data in the third data set with the randomly extracted data in the second sample library according to a preset mode to generate a fourth data set.
2. The method of claim 1, wherein:
the first, second, third, and fourth data sets are sets of image data; wherein the content of the first and second substances,
the first data set is a collection of image data conforming to a predetermined characteristic;
the second data set is a set of image data that does not contain the predetermined feature;
the third data set is a set of image data to be enhanced;
the fourth data set is a set of images that have undergone image enhancement processing.
3. The method of claim 1 or 2, wherein:
the first data set is a set of images of the parts when the parts have defects;
the second data set is a set of images of the parts when the parts have no defects;
the third data set refers to a set of part images of the image to be enhanced;
the fourth data set refers to a set of part images after image enhancement processing.
4. The method of claim 3, wherein:
the component part comprises an automobile component part.
5. The method of claim 3, wherein:
when the part meets the preset characteristics, the part is defective;
wherein the defect at least comprises one of the following: flash, scratch, stain, deformation, impurities;
the preset characteristics are set according to the process requirements of the parts;
the predetermined characteristic is extensible.
6. The method of claim 3, wherein:
the random enhancement comprises at least one of scale scaling, rotation transformation and brightness adjustment;
the algorithm adopted by the random enhancement is pre-designated or randomly selected;
the algorithm adopted by the random enhancement can be redesigned according to the process requirements.
7. The method of any of claims 4-6, wherein:
after the merging, the defects exist only in the areas where the defects can appear;
wherein the selection of the regions is random.
8. The method of claim 3, wherein:
observing an obtained image from a certain direction outside the part, wherein the image comprises a symmetrical characteristic or an asymmetrical characteristic;
the random enhancement means that the observed image or the local area of the observed image is randomly rotated by an angle;
when the observed image includes an annotation, the randomly enhanced processing includes processing of the annotation.
9. The method of claim 8, wherein:
the symmetrical characteristic is a circle, a circular ring or other similar shapes and lines;
when the first sample library is generated, the following rules are not limited to the gray level binarization defect extraction and/or sobel operator contour extraction rules.
10. The method of claim 9, wherein:
the first data set is an empty set.
11. The method of claim 9 or 10, wherein:
the component parts are not limited to plastic or metal parts.
12. A defect detection apparatus, comprising:
the device comprises a feature extraction unit (100), a random enhancement unit (200), a data extraction unit (300) and a random mapping unit (400), wherein;
the method comprises the following steps that a characteristic extraction unit (100) acquires data which are in accordance with preset characteristics in a first data set, and generates a first sample library;
a random enhancement unit (200) performs random enhancement on the first sample library to generate a second sample library;
the data extraction unit (300) extracts data in the first data set or the second data set to generate a third data set;
a random mapping unit (400) combines the data in the third data set with the randomly drawn data in the second sample library in a predetermined manner to generate a fourth data set.
13. The apparatus of claim 12, further comprising:
a microprocessor and a memory electrically connected, wherein;
the memory is used for storing a computer program; the computer program is adapted to carry out the method of any one of claims 1 to 11 when executed by a microprocessor.
14. A computer-readable storage medium, comprising:
a storage medium body for storing a computer program; the computer program is adapted to carry out the method of any one of claims 1 to 11 when executed by a microprocessor.
15. A vehicle, comprising:
an apparatus according to claims 12-13 or a storage medium according to claim 14.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110578233.XA CN113393388A (en) | 2021-05-26 | 2021-05-26 | Image enhancement method, device adopting same, storage medium and vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110578233.XA CN113393388A (en) | 2021-05-26 | 2021-05-26 | Image enhancement method, device adopting same, storage medium and vehicle |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113393388A true CN113393388A (en) | 2021-09-14 |
Family
ID=77619227
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110578233.XA Pending CN113393388A (en) | 2021-05-26 | 2021-05-26 | Image enhancement method, device adopting same, storage medium and vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113393388A (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109584206A (en) * | 2018-10-19 | 2019-04-05 | 中国科学院自动化研究所 | The synthetic method of the training sample of neural network in piece surface Defect Detection |
| CN111223058A (en) * | 2019-12-27 | 2020-06-02 | 杭州雄迈集成电路技术股份有限公司 | Image enhancement method |
| CN111428664A (en) * | 2020-03-30 | 2020-07-17 | 厦门瑞为信息技术有限公司 | Real-time multi-person posture estimation method based on artificial intelligence deep learning technology for computer vision |
| CN111428807A (en) * | 2020-04-03 | 2020-07-17 | 桂林电子科技大学 | Image processing method and computer-readable storage medium |
| CN111881720A (en) * | 2020-06-09 | 2020-11-03 | 山东大学 | Data automatic enhancement expansion method, data automatic enhancement identification method and data automatic enhancement expansion system for deep learning |
| CN111951195A (en) * | 2020-07-08 | 2020-11-17 | 华为技术有限公司 | Image enhancement method and device |
| CN112365423A (en) * | 2020-11-23 | 2021-02-12 | 腾讯科技（深圳）有限公司 | Image data enhancement method, device, medium and equipment |
| CN112749746A (en) * | 2021-01-12 | 2021-05-04 | 云南电网有限责任公司电力科学研究院 | Method, system and device for iteratively updating defect sample |
-
2021
- 2021-05-26 CN CN202110578233.XA patent/CN113393388A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109584206A (en) * | 2018-10-19 | 2019-04-05 | 中国科学院自动化研究所 | The synthetic method of the training sample of neural network in piece surface Defect Detection |
| CN111223058A (en) * | 2019-12-27 | 2020-06-02 | 杭州雄迈集成电路技术股份有限公司 | Image enhancement method |
| CN111428664A (en) * | 2020-03-30 | 2020-07-17 | 厦门瑞为信息技术有限公司 | Real-time multi-person posture estimation method based on artificial intelligence deep learning technology for computer vision |
| CN111428807A (en) * | 2020-04-03 | 2020-07-17 | 桂林电子科技大学 | Image processing method and computer-readable storage medium |
| CN111881720A (en) * | 2020-06-09 | 2020-11-03 | 山东大学 | Data automatic enhancement expansion method, data automatic enhancement identification method and data automatic enhancement expansion system for deep learning |
| CN111951195A (en) * | 2020-07-08 | 2020-11-17 | 华为技术有限公司 | Image enhancement method and device |
| CN112365423A (en) * | 2020-11-23 | 2021-02-12 | 腾讯科技（深圳）有限公司 | Image data enhancement method, device, medium and equipment |
| CN112749746A (en) * | 2021-01-12 | 2021-05-04 | 云南电网有限责任公司电力科学研究院 | Method, system and device for iteratively updating defect sample |
Non-Patent Citations (3)
| Title |
|---|
| LIMING SUN等: "A new infrared image enhancement algorithm", 2017 29TH CHINESE CONTROL AND DECISION CONFERENCE (CCDC), pages 421 - 424 * |
| 王帅: "基于PCNN的多聚焦图像融合算法研究", 中国优秀博硕士学位论文全文数据库(硕士), vol. 2018, no. 03 * |
| 邵桂芳等: "一种基于同类融合的工业数据扩充算法", 厦门大学学报(自然科学版), vol. 60, no. 01, 28 January 2021 (2021-01-28), pages 80 - 90 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109978839B (en) | Method for detecting wafer low-texture defects | |
| CN107228860B (en) | Gear defect detection method based on image rotation period characteristics | |
| CN107808161B (en) | Underwater target identification method based on optical vision | |
| CN111553858B (en) | Image restoration method and system based on generation countermeasure network and application thereof | |
| CN115641332B (en) | Method, device, medium and equipment for detecting product edge appearance defects | |
| Shin et al. | Data augmentation method of object detection for deep learning in maritime image | |
| CN111476758B (en) | Defect detection method and device for AMOLED display screen, computer equipment and storage medium | |
| US11823457B2 (en) | Image recognition method and apparatus, based on context representation, and computer-readable storage medium | |
| CN110599453A (en) | Panel defect detection method and device based on image fusion and equipment terminal | |
| CN113409234A (en) | Minimum supervision Automatic Inspection (AI) of wafers supported by Convolutional Neural Network (CNN) algorithm | |
| CN114170227A (en) | Product surface defect detection method, device, equipment and storage medium | |
| CN111582004A (en) | Target area segmentation method and device in ground image | |
| CN113642576A (en) | Method and device for generating training image set in target detection and semantic segmentation task | |
| Kim et al. | Deep illumination-aware dehazing with low-light and detail enhancement | |
| CN115797314B (en) | Method, system, equipment and storage medium for detecting surface defects of parts | |
| CN113393388A (en) | Image enhancement method, device adopting same, storage medium and vehicle | |
| CN116883987A (en) | Pointer instrument reading identification method for unmanned inspection of transformer substation | |
| CN116824135A (en) | Atmospheric natural environment test industrial product identification and segmentation method based on machine vision | |
| KR20230116985A (en) | System and method for detecting pit defects on film | |
| CN115578362A (en) | Defect detection method and device for electrode coating, electronic device and medium | |
| CN113610838A (en) | Bolt defect data set expansion method | |
| CN113989814A (en) | Image generation method and device, computer equipment and storage medium | |
| CN114240837A (en) | Welding seam positioning method, device, equipment and storage medium | |
| CN112652004B (en) | Image processing method, device, equipment and medium | |
| CN117576105B (en) | Pipeline production control method and system based on artificial intelligence |
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 |