CN111062408A - Fuzzy license plate image super-resolution reconstruction method based on deep learning - Google Patents
Fuzzy license plate image super-resolution reconstruction method based on deep learning Download PDFInfo
- Publication number
- CN111062408A CN111062408A CN201911011865.7A CN201911011865A CN111062408A CN 111062408 A CN111062408 A CN 111062408A CN 201911011865 A CN201911011865 A CN 201911011865A CN 111062408 A CN111062408 A CN 111062408A
- Authority
- CN
- China
- Prior art keywords
- license plate
- plate character
- character image
- image
- fuzzy
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000013135 deep learning Methods 0.000 title claims abstract description 22
- 238000012545 processing Methods 0.000 claims abstract description 36
- 238000012549 training Methods 0.000 claims abstract description 9
- 238000007781 pre-processing Methods 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 14
- 238000005070 sampling Methods 0.000 claims description 6
- 230000000903 blocking effect Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 4
- 238000012216 screening Methods 0.000 claims description 3
- 238000000638 solvent extraction Methods 0.000 claims description 3
- 238000011840 criminal investigation Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4038—Scaling the whole image or part thereof for image mosaicing, i.e. plane images composed of plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4053—Super resolution, i.e. output image resolution higher than sensor resolution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration by the use of more than one image, e.g. averaging, subtraction
-
- G06T5/70—
-
- G06T5/77—
-
- 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
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- 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
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
- G06V20/625—License plates
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
Abstract
The invention discloses a deep learning-based fuzzy license plate image super-resolution reconstruction method, and relates to the technical field of video and image processing. The method comprises the following steps: preprocessing a license plate character image, and performing fuzzy processing on a clear license plate character image in a database; training a classifier by taking the license plate character image subjected to fuzzy processing as a sample; classifying license plate character images to be recognized by using a trained classifier and taking the first N recognition results with the maximum prediction probability; for each recognition result, taking the similarity between the license plate character image after fuzzy processing and the license plate character image to be recognized as a weight, and carrying out weighted summation on the corresponding clear license plate character image to obtain a pre-reconstruction block; taking the maximum first N prediction probabilities as weights, and carrying out weighted summation on the corresponding pre-reconstruction blocks to obtain final reconstruction blocks; and splicing the reconstructed blocks according to the position relation. The invention reconstructs the fuzzy license plate characters and accurately restores the original face of the license plate characters.
Description
Technical Field
The invention relates to the technical field of video and image processing, in particular to a fuzzy license plate image super-resolution reconstruction method based on deep learning.
Background
The vehicle information has important significance in the fields of public safety systems and criminal investigation. Due to the shooting angle, distance conditions, equipment factors, environmental factors and the like, the license plate in the monitored video is fuzzy, and the license plate is difficult to distinguish and distinguish by naked eyes when the license plate is fuzzy. In the meantime, although the vehicle character recognition system can recognize characters and then push a final recognition result, under the conditions of severe data acquisition environment and complex scene, the vehicle character recognition result has a high error probability, for example, in a crime evidence obtaining link in the criminal investigation field, the license plate character information plays an important role in vehicle tracking and identity discrimination, but the license plate character information in the scene is generally fuzzy, and the traditional license plate character recognition result has errors, or the characters in the scene cannot be recognized accurately.
Disclosure of Invention
The invention aims to provide a method for reconstructing a super-resolution fuzzy license plate image based on deep learning, which is used for reconstructing fuzzy license plate characters and accurately recovering the original faces of the license plate characters.
In order to achieve the purpose, the invention provides the following technical scheme:
a super-resolution reconstruction method of a blurred license plate image based on deep learning is characterized by comprising the following steps:
s1, preprocessing the license plate character image, and performing fuzzy processing on the clear license plate character image in the database;
s2, training a classifier by taking the license plate character image after fuzzy processing as a sample;
s3, classifying the license plate character images to be recognized by using a trained classifier and taking the first N recognition results with the maximum prediction probability;
s4, aiming at each recognition result, taking the similarity between the license plate character image after fuzzy processing and the license plate character image to be recognized as a weight, and carrying out weighted summation on the corresponding clear license plate character image to obtain a pre-reconstruction block;
s5, taking the maximum first N prediction probabilities as weights, and carrying out weighted summation on the corresponding pre-reconstruction blocks to obtain final reconstruction blocks;
and S6, splicing the reconstructed blocks according to the position relation.
Further, the blurring processing is one or more of motion blurring, random noise, mosaic and defocusing blurring.
Further, the classifier adopts a GoogleNet deep learning network for classifier training.
Further, the step of S4 is as follows:
s41, aiming at each recognition result, constructing a clear image sample set by taking the corresponding clear license plate character image in the database, and constructing a fuzzy image sample set by the corresponding fuzzy license plate character image;
s42, partitioning the license plate character image to be recognized;
s43, performing the blocking operation such as S42 on the clear license plate character image and the license plate character image after the fuzzy processing;
s44, solving the weight of each blurred license plate character image block at the same position;
and S45, carrying out weighted summation according to the weight and the corresponding clear license plate character image blocks to obtain pre-reconstruction blocks.
Further, the step S41 includes size normalization, which is to perform down-sampling or up-sampling on the clear license plate character image and the license plate character image after the blur processing, so that the sizes of the clear license plate character image and the license plate character image after the blur processing are the same as the size of the license plate character image to be recognized.
Further, a sample screening operation is also included before S44;
calculating the similarity between the blocks of the license plate character images to be recognized at the same position and the blocks of the license plate character images after fuzzy processing, and taking the blocks of the first K fuzzy processed license plate character images with the highest similarity as samples of the solving weight in S4.
Further, K is 900.
Further, in S44, the weight is calculated as follows:
s441, reducing the blocks of the license plate character image to be recognized into a one-dimensional matrix Blur (i, j)' according to columns;
s442, reducing the dimension of the blocks of the fuzzy license plate character images into a one-dimensional matrix according to columns, and splicing the blocks of the fuzzy license plate character images into a set matrix { DB (database data base) according to columnsproc(i,j)}；
S443, weight calculation:
S＝Blur(i,j)′*OT-{DBproc(i,j)}
Z＝ST*S
w＝(Z-1*O)/(OT*Z-1*O)
wherein O is a full 1 matrix of C1; w is a vector of 1 × C, w ═ wc}，wcAnd representing the weight corresponding to the block of the c-th fuzzy processed license plate character image corresponding to the position in the database.
Further, the weighting and summing method in S45 is as follows:
wherein, HRVehichern(i, j) is the pre-reconstructed block located at position (i, j) for the nth recognition result, DBhr(i,j)cAnd (4) dividing the c clear license plate character image corresponding to the position (i, j) in the database into blocks.
Compared with the prior art, the invention has the beneficial effects that: aiming at license plate images in low-resolution videos and images, various fuzzy pretreatments are carried out according to clear license plate character images, then a deep learning method is used for training a license plate character classifier, the deep learning method can more fully extract character image characteristics, and the problem that the license plate in the low-resolution videos cannot be identified due to fuzzy is effectively solved; then, according to the result predicted by the classifier, carrying out similarity reconstruction based on image blocking; and finally, performing block fusion to complete super-resolution reconstruction. The license plate character image reconstructed by the method has complete and continuous texture and can well restore the original appearance.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Fig. 2 is a cut clear image of license plate characters according to an embodiment of the present invention.
Fig. 3 is a diagram illustrating a plurality of license plate character images obtained by blurring the same license plate character image according to an embodiment of the present invention.
Fig. 4 is a blurred license plate character image to be recognized according to an embodiment of the present invention.
Fig. 5 shows the final reconstruction result according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention are clearly and completely described below, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, the present invention provides a method for reconstructing a super-resolution blurred license plate image based on deep learning, which includes the following steps:
and S1, preprocessing the license plate character image, and performing fuzzy processing on the clear license plate character image in the database.
And (3) carrying out fuzzy processing of a simulation environment on the clear license plate character image shown in the figure 2 in the database to obtain a corresponding license plate character image after fuzzy processing shown in the figure 3. The fuzzy processing adopts a conventional algorithm, and preferably adopts one or more of motion blur, random noise, mosaic and defocus blur processing. It is worth mentioning that, for the same license plate character such as the letter B, a plurality of clear license plate character images are in the database, and after the blurring processing, each clear license plate character image can obtain a plurality of corresponding blurred license plate character images.
And S2, training a classifier by taking the license plate character image subjected to the fuzzy processing as a sample.
And performing deep learning training on all license plate character images in the database, including clear license plate character images and license plate character images after fuzzy processing, preferably performing classifier training by adopting a GoogleNet deep learning network to obtain a network model parameter model of the classifier.
And S3, classifying the license plate character images to be recognized by using a trained classifier and taking the first N recognition results with the maximum prediction probability.
And cutting the license plate image to be recognized to obtain a license plate character image vehiChar to be recognized as shown in fig. 4. And identifying the license plate character image vehiChar to be identified by using a trained classifier to obtain corresponding prediction probability, sequencing the prediction probabilities from large to small, taking the previous N identification result, preferably N is 5, and obtaining the following corresponding results:
Result＝{Res1:Rate1,Res2:Rate2,...,Res5:Rate5}
wherein Resn:RatenRepresenting the nth recognition result; resnIs the corresponding character category; ratenFor corresponding to the character type ResnN is a natural number and N is more than or equal to 1 and less than or equal to N.
And S4, taking the similarity between the license plate character image after the fuzzy processing and the license plate character image to be recognized as a weight according to each recognition result, and carrying out weighted summation on the corresponding clear license plate character image to obtain a pre-reconstruction block.
Taking the 1 st recognition result as an example, the specific steps of S4 are as follows:
s41, aiming at the character type Res1The corresponding character type Res in the database1Constructing a clear image sample set HRChars for the samples from the step S42 to the step S45 by using the corresponding clear license plate character image; similarly, a blurred image sample set procChars is constructed for the corresponding blurred license plate character image, and is used as a sample from step S42 to step S45. It is worth mentioning that in order to make the sizes of the subsequent block images consistent, the step further comprises size normalization, wherein clear license plate character images in the sample set HRChars and the sample set are combinedDown-sampling or up-sampling the license plate character image after fuzzy processing in procChars to enable the size of the clear license plate character image and the license plate character image after fuzzy processing to be the same as the size of the license plate character image vehiChar to be recognized, and respectively obtaining the clear license plate character image DB after size normalizationhrAnd the license plate character image DB after fuzzy processingproc。
And S42, performing equal-distance blocking on the license plate character image vehiChar to be recognized. Preferably, the license plate character image vehiChar to be recognized is equally divided into 32 blocks by 8 × 4. Wherein, the image blocks of the ith row and the jth column are denoted as Blur (i, j).
S43, clear license plate character image DBhrAnd the license plate character image DB after fuzzy processingprocThe blocking operation is performed as S42 to obtain DBhr(i, j) and DBproc(i,j)。DBhr(i, j) is the image block of the ith row and the jth column of the clear license plate character image; DBprocAnd (i, j) is the image block of the ith row and the jth column of the license plate character image after the fuzzy processing.
S44, solving the weight of each blurred license plate character image block at the same position;
the weight calculation method comprises the following steps:
s441, reducing the blocks Blur (i, j) of the license plate character image to be recognized into a one-dimensional matrix Blur (i, j)' according to columns; taking the 2 × 3 matrix as an example, the 6 × 1 matrix is obtained by data concatenation column by column from left to right, and the corresponding conversion mode is as follows:
s442, partitioning DB of the fuzzy-processed license plate character imageproc(i, j) as shown in S441, reducing dimension into a one-dimensional matrix according to columns, and splicing blocks of the license plate character image after fuzzy processing into an aggregate matrix { DB (database) } according to columnsproc(i, j) }; the splicing method is as follows:
s443, weight calculation:
S＝Blur(i,j)′*OT-{DBproc(i,j)}
Z＝ST*S
w＝(Z-1*O)/(OT*Z-1*O)
wherein O is a full 1 matrix of C1; w is a vector of 1 × C, w ═ wc}，wcA block DB representing the c-th blurred license plate character image corresponding to the position (i, j) in the databaseproc(i, j) corresponding weights. It is worth mentioning that w in wcAnd the block DB of each blurred license plate character image in S442proc(i, j) correspond in the order of column concatenation.
Preferably, when the number of samples in the database is too large, in order to reduce the calculation amount, a sample screening operation is further included before the step S44;
specifically, the block Blur (i, j) of the license plate character image to be recognized at the same position (i, j) and the block DB of each license plate character image after fuzzy processing are calculatedproc(ii) the degree of similarity of (i, j),
wherein Blur (i, j) (m, n) represents the pixel values of the m-th row and the n-th column of the block Blur (i, j) of the license plate character image to be recognized; DBproc(i, j) (m, n) block DB for the license plate character image after the blurring processproc(i, j) pixel values of row m, column n; the blocks Blur (i, j) of the license plate character image to be recognized and the block DB of the license plate character image after fuzzy processingproc(i, j) are matrices of M N.
And taking the blocks of the first K blurred license plate character images with the highest similarity as samples for solving the weight in the S4. K is preferably 900.
S45, according to the weight wcAnd the corresponding clear license plate character image block DBhrAnd (i, j) carrying out weighted summation to obtain the pre-reconstruction blocks. Weighted solutionAnd the method is as follows:
wherein, HRVehichern(i, j) is the pre-reconstructed block located at position (i, j) for the nth recognition result, where n is 1 corresponding to the 1 st recognition result. DBhr(i,j)cFor the c-th clear license plate character image block corresponding to the position (i, j) in the database, and solving the weight w in S443cBlock DB of license plate character image after time-fuzzy processingproc(i, j) correspond. For example, as shown in step S1, since any clear license plate character image corresponds to a plurality of blurred license plate character images, the c-th clear license plate character image DBhr(i,j)cIn other words, the license plate character image DB corresponding to a plurality of fuzzy processed license plate character imagesproc(i,j)。
For the rest of the recognition results, the contents of the step S4 are repeated to respectively solve the rest of the pre-reconstruction blocks HRVehichar at the position (i, j)n(i,j)。
And S5, fusing final characters, wherein the maximum first N prediction probabilities are used as weights, and N is 5. Carrying out weighted summation on the corresponding pre-reconstruction blocks to obtain final reconstruction blocks;
wherein, the vehiCharFinal (i, j) is the final reconstructed block corresponding to the position (i, j).
And S6, splicing the reconstruction blocks according to the positions (i, j) to obtain a final reconstruction result shown in FIG. 5.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.
Claims (9)
1. A super-resolution reconstruction method of a blurred license plate image based on deep learning is characterized by comprising the following steps:
s1, preprocessing the license plate character image, and performing fuzzy processing on the clear license plate character image in the database;
s2, training a classifier by taking the license plate character image after fuzzy processing as a sample;
s3, classifying the license plate character images to be recognized by using a trained classifier and taking the first N recognition results with the maximum prediction probability;
s4, aiming at each recognition result, taking the similarity between the license plate character image after fuzzy processing and the license plate character image to be recognized as a weight, and carrying out weighted summation on the corresponding clear license plate character image to obtain a pre-reconstruction block;
s5, taking the maximum first N prediction probabilities as weights, and carrying out weighted summation on the corresponding pre-reconstruction blocks to obtain final reconstruction blocks;
and S6, splicing the reconstructed blocks according to the position relation.
2. The method for super-resolution reconstruction of the blurred license plate image based on deep learning of claim 1, wherein the blurring process is one or more of motion blur, random noise, mosaic and defocus blur.
3. The deep learning-based fuzzy license plate image super-resolution reconstruction method of claim 1, wherein the classifier adopts a GoogleNet deep learning network for classifier training.
4. The method for super-resolution reconstruction of the blurred license plate image based on deep learning of claim 1, wherein the step of S4 is as follows:
s41, aiming at each recognition result, constructing a clear image sample set by taking the corresponding clear license plate character image in the database, and constructing a fuzzy image sample set by the corresponding fuzzy license plate character image;
s42, partitioning the license plate character image to be recognized;
s43, performing the blocking operation such as S42 on the clear license plate character image and the license plate character image after the fuzzy processing;
s44, solving the weight of each blurred license plate character image block at the same position;
and S45, carrying out weighted summation according to the weight and the corresponding clear license plate character image blocks to obtain pre-reconstruction blocks.
5. The method for super-resolution reconstruction of the blurred license plate image based on the deep learning of claim 4, wherein the step S41 further comprises size normalization, and the step S carries out down-sampling or up-sampling on the sharp license plate character image and the blurred license plate character image, so that the sizes of the sharp license plate character image and the blurred license plate character image are the same as the size of the license plate character image to be recognized.
6. The method for super-resolution reconstruction of the blurred license plate image based on deep learning of claim 4, wherein a sample screening operation is further included before the step S44;
calculating the similarity between the blocks of the license plate character images to be recognized at the same position and the blocks of the license plate character images after fuzzy processing, and taking the blocks of the first K fuzzy processed license plate character images with the highest similarity as samples of the solving weight in S4.
7. The method for super-resolution reconstruction of the blurred license plate image based on deep learning of claim 6, wherein K is 900.
8. The method for super-resolution reconstruction of the blurred license plate image based on deep learning of claim 4, wherein in S44, the weight is calculated as follows:
s441, reducing the blocks of the license plate character image to be recognized into a one-dimensional matrix Blur (i, j)' according to columns;
s442, reducing the dimension of the blocks of the fuzzy license plate character images into a one-dimensional matrix according to columns, and splicing the blocks of the fuzzy license plate character images into a set matrix { DB (database data base) according to columnsproc(i,j)}；
S443, weight calculation:
S＝Blur(i,j)′*OT-{DBproc(i,j)}
Z＝ST*S
w＝(Z-1*O)/(OT*Z-1*O)
wherein O is a full 1 matrix of C1; w is a vector of 1 × C, w ═ wc}，wcAnd representing the weight corresponding to the block of the c-th fuzzy processed license plate character image corresponding to the position in the database.
9. The method for super-resolution reconstruction of the blurred license plate image based on deep learning of claim 8, wherein the weighting and summing method in S45 is as follows:
wherein, HRVehichern(i, j) is the pre-reconstructed block located at position (i, j) for the nth recognition result, DBhr(i,j)cAnd (4) dividing the c clear license plate character image corresponding to the position (i, j) in the database into blocks.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911011865.7A CN111062408B (en) | 2019-10-23 | 2019-10-23 | Fuzzy license plate image super-resolution reconstruction method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911011865.7A CN111062408B (en) | 2019-10-23 | 2019-10-23 | Fuzzy license plate image super-resolution reconstruction method based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111062408A true CN111062408A (en) | 2020-04-24 |
| CN111062408B CN111062408B (en) | 2023-04-18 |
Family
ID=70297668
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911011865.7A Active CN111062408B (en) | 2019-10-23 | 2019-10-23 | Fuzzy license plate image super-resolution reconstruction method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111062408B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113792739A (en) * | 2021-08-25 | 2021-12-14 | 电子科技大学 | Universal license plate text recognition method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101477684A (en) * | 2008-12-11 | 2009-07-08 | 西安交通大学 | Process for reconstructing human face image super-resolution by position image block |
| CN103903236A (en) * | 2014-03-10 | 2014-07-02 | 北京信息科技大学 | Method and device for reconstructing super-resolution facial image |
| CN107480772A (en) * | 2017-08-08 | 2017-12-15 | 浙江大学 | A kind of car plate super-resolution processing method and system based on deep learning |
| US20190206026A1 (en) * | 2018-01-02 | 2019-07-04 | Google Llc | Frame-Recurrent Video Super-Resolution |
| CN110223231A (en) * | 2019-06-06 | 2019-09-10 | 天津工业大学 | A kind of rapid super-resolution algorithm for reconstructing of noisy image |
-
2019
- 2019-10-23 CN CN201911011865.7A patent/CN111062408B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101477684A (en) * | 2008-12-11 | 2009-07-08 | 西安交通大学 | Process for reconstructing human face image super-resolution by position image block |
| CN103903236A (en) * | 2014-03-10 | 2014-07-02 | 北京信息科技大学 | Method and device for reconstructing super-resolution facial image |
| CN107480772A (en) * | 2017-08-08 | 2017-12-15 | 浙江大学 | A kind of car plate super-resolution processing method and system based on deep learning |
| US20190206026A1 (en) * | 2018-01-02 | 2019-07-04 | Google Llc | Frame-Recurrent Video Super-Resolution |
| CN110223231A (en) * | 2019-06-06 | 2019-09-10 | 天津工业大学 | A kind of rapid super-resolution algorithm for reconstructing of noisy image |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113792739A (en) * | 2021-08-25 | 2021-12-14 | 电子科技大学 | Universal license plate text recognition method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111062408B (en) | 2023-04-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111444881B (en) | Fake face video detection method and device | |
| CN108256562B (en) | Salient target detection method and system based on weak supervision time-space cascade neural network | |
| CN108805015B (en) | Crowd abnormity detection method for weighted convolution self-coding long-short term memory network | |
| CN108564127B (en) | Image conversion method, image conversion device, computer equipment and storage medium | |
| CN110110601A (en) | Video pedestrian weight recognizer and device based on multi-space attention model | |
| CN111079655B (en) | Method for recognizing human body behaviors in video based on fusion neural network | |
| CN111738054B (en) | Behavior anomaly detection method based on space-time self-encoder network and space-time CNN | |
| CN111985374A (en) | Face positioning method and device, electronic equipment and storage medium | |
| CN113255659A (en) | License plate correction detection and identification method based on MSAFF-yolk 3 | |
| CN116910752B (en) | Malicious code detection method based on big data | |
| CN111626134A (en) | Dense crowd counting method, system and terminal based on hidden density distribution | |
| CN113205002A (en) | Low-definition face recognition method, device, equipment and medium for unlimited video monitoring | |
| CN112149526A (en) | Lane line detection method and system based on long-distance information fusion | |
| CN111062408B (en) | Fuzzy license plate image super-resolution reconstruction method based on deep learning | |
| CN115293966A (en) | Face image reconstruction method and device and storage medium | |
| Nayak et al. | Video anomaly detection using convolutional spatiotemporal autoencoder | |
| CN114202473A (en) | Image restoration method and device based on multi-scale features and attention mechanism | |
| CN111144220B (en) | Personnel detection method, device, equipment and medium suitable for big data | |
| Li et al. | A new qr code recognition method using deblurring and modified local adaptive thresholding techniques | |
| CN116934725A (en) | Method for detecting sealing performance of aluminum foil seal based on unsupervised learning | |
| CN112668378A (en) | Facial expression recognition method based on combination of image fusion and convolutional neural network | |
| CN116092179A (en) | Improved Yolox fall detection system | |
| CN116110095A (en) | Training method of face filtering model, face recognition method and device | |
| CN115797970A (en) | Dense pedestrian target detection method and system based on YOLOv5 model | |
| CN111931587B (en) | Video anomaly detection method based on interpretable space-time self-encoder |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20230320 Address after: 310000 Room 401, building 2, No.16, Zhuantang science and technology economic block, Xihu District, Hangzhou City, Zhejiang Province Applicant after: Hangzhou yunqi smart Vision Technology Co.,Ltd. Address before: 310000 room 279, building 6, No. 16, Zhuantang science and technology economic block, Zhuantang street, Xihu District, Hangzhou City, Zhejiang Province Applicant before: Smart vision (Hangzhou) Technology Development Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |