CN111325670A - Data enhancement method and device and electronic equipment - Google Patents
Data enhancement method and device and electronic equipment Download PDFInfo
- Publication number
- CN111325670A CN111325670A CN201811526923.5A CN201811526923A CN111325670A CN 111325670 A CN111325670 A CN 111325670A CN 201811526923 A CN201811526923 A CN 201811526923A CN 111325670 A CN111325670 A CN 111325670A
- Authority
- CN
- China
- Prior art keywords
- image
- enhanced
- adjusting
- oblique
- images
- 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 54
- 238000005286 illumination Methods 0.000 claims abstract description 56
- 230000009466 transformation Effects 0.000 claims description 99
- 239000011159 matrix material Substances 0.000 claims description 46
- 238000012937 correction Methods 0.000 claims description 26
- 238000004891 communication Methods 0.000 claims description 17
- 238000004422 calculation algorithm Methods 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 8
- 230000000694 effects Effects 0.000 abstract description 19
- 230000002708 enhancing effect Effects 0.000 abstract 1
- 230000008569 process Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 5
- 230000001131 transforming effect Effects 0.000 description 5
- 230000007423 decrease Effects 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000013434 data augmentation Methods 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000009336 multiple cropping Methods 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
Images
Classifications
-
- G06T5/80—
-
- G06T5/90—
Abstract
The embodiment of the invention provides a method and a device for enhancing data and electronic equipment, wherein the method comprises the following steps: acquiring an image to be enhanced; adjusting the inclination angle of the image to be enhanced to obtain an inclined image; and adjusting the gray value of the oblique image to obtain the oblique image under different illumination conditions as a generated enhanced image. By applying the embodiment of the invention, the image is adjusted from two aspects of the image inclination angle and the gray value, the two aspects of adjustment do not influence the content of the image, the availability of the obtained enhanced image is higher, and the data enhancement effect is improved.
Description
Technical Field
The present invention relates to the field of data processing technologies, and in particular, to a data enhancement method and apparatus, and an electronic device.
Background
With the development of artificial intelligence, machine learning techniques are used more and more widely. In some cases, the recognition model may be trained based on the sample images through a machine learning algorithm. For example, a character recognition model may be trained based on sample images containing characters; the character recognition model can be used for recognizing various images containing characters, such as license plate images. For another example, a face recognition model may be obtained by training based on a sample image including a face region, and the face image may be recognized by using the face recognition model.
In order to improve the recognition accuracy of the recognition model, a large number of sample images are generally collected. In the prior art, most of the existing solutions collect some sample images, and then perform an expansion process, or data enhancement, on the sample images to obtain more sample images.
Related data enhancement schemes generally include: and turning the collected sample image, for example, horizontally turning, vertically turning or turning at other angles, wherein the turned image is the image after the enhancement processing. However, the usability of the inverted sample character image is not high, for example, the character 6 is turned over to become the character 9, and the character n is turned over to become the character u, which is not in accordance with the intention. It can be seen that the data enhancement scheme has a poor enhancement effect.
Disclosure of Invention
The embodiment of the invention aims to provide a data enhancement method, a data enhancement device and electronic equipment so as to improve the data enhancement effect.
In order to achieve the above object, an embodiment of the present invention discloses a data enhancement method, including:
acquiring an image to be enhanced;
adjusting the inclination angle of the image to be enhanced to obtain an inclined image;
and adjusting the gray value of the oblique image to obtain the oblique image under different illumination conditions as a generated enhanced image.
Optionally, the adjusting the inclination angle of the image to be enhanced to obtain an inclined image includes:
and adjusting the inclination angle of the image to be enhanced by carrying out perspective transformation on the image to be enhanced to obtain an inclined image.
Optionally, the obtaining of the oblique image by performing perspective transformation on the image to be enhanced and adjusting the oblique angle of the image to be enhanced includes:
according to the left inclination transformation matrix, adjusting the image to be enhanced to incline to the left to obtain a left inclined image;
and adjusting the image to be enhanced to be inclined to the right according to the right inclination transformation matrix to obtain a right inclined image.
Optionally, the adjusting the gray value of the oblique image to obtain the oblique image under different illumination conditions as the generated enhanced image includes:
and carrying out nonlinear adjustment on the gray value in the oblique image by utilizing a gamma correction algorithm to obtain the oblique image under different illumination conditions as a generated enhanced image.
Optionally, after the acquiring the image to be enhanced, the method further includes:
determining an identification target area in the image to be enhanced;
taking the identification target area as a reference, and cutting the image to be enhanced for multiple times to obtain multiple cut images;
the adjusting the inclination angle of the image to be enhanced to obtain an inclined image comprises:
and adjusting the inclination angles of the plurality of cut images respectively to obtain a plurality of inclined images.
Optionally, the performing, with the recognition target area as a reference, multiple cropping on the image to be enhanced to obtain multiple cropped images includes:
adding different boundary disturbance parameters to the boundary of the identification target area respectively to obtain different identification target frames;
and respectively cutting out each recognition target frame from the image to be enhanced to be used as a cut image.
Optionally, the image to be enhanced is a license plate image;
after the obtaining of the oblique images under different illumination conditions by adjusting the gray-scale values of the oblique images as the generated enhanced images, the method further includes:
and adding the generated enhanced image to a sample training set of a license plate recognition model.
In order to achieve the above object, an embodiment of the present invention further discloses a data enhancement apparatus, including:
the image acquisition module is used for acquiring an image to be enhanced;
the angle adjusting module is used for adjusting the inclination angle of the image to be enhanced to obtain an inclined image;
and the gray level adjusting module is used for adjusting the gray level value of the oblique image to obtain the oblique image under different illumination conditions as the generated enhanced image.
Optionally, the angle adjusting module is specifically configured to adjust an inclination angle of the image to be enhanced by performing perspective transformation on the image to be enhanced, so as to obtain an inclined image.
Optionally, the angle adjusting module includes:
the left inclination transformation submodule is used for adjusting the image to be enhanced to incline to the left according to a left inclination transformation matrix to obtain a left inclination image;
and the right inclination transformation submodule is used for adjusting the image to be enhanced to be inclined to the right according to the right inclination transformation matrix to obtain a right inclined image.
Optionally, the gray scale adjustment module is specifically configured to perform nonlinear adjustment on the gray scale value in the oblique image by using a gamma correction algorithm to obtain the oblique image under different illumination conditions, and the oblique image is used as the generated enhanced image.
Optionally, the apparatus further comprises:
the region determining module is used for determining a recognition target region in the image to be enhanced after the image to be enhanced is obtained;
the cutting module is used for cutting the image to be enhanced for multiple times by taking the identification target area as a reference to obtain multiple cut images;
the angle adjusting module is specifically configured to adjust the tilt angles of the plurality of cut images respectively to obtain a plurality of tilted images.
Optionally, the cutting module includes:
the boundary disturbance adding submodule is used for respectively adding different boundary disturbance parameters to the boundary of the identification target area to obtain different identification target frames;
and the cutting submodule is used for respectively cutting each recognition target frame from the image to be enhanced to be used as a cut image.
Optionally, if the image to be enhanced is a license plate image, the apparatus further includes:
and the image adding module is used for adjusting the gray value of the oblique image to obtain the oblique image under different illumination conditions, and adding the generated enhanced image to a sample training set of the license plate recognition model after the oblique image is used as the generated enhanced image.
In order to achieve the above object, an embodiment of the present invention further discloses an electronic device, where the device includes: the system comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory are communicated with each other through the communication bus;
the memory is used for storing a computer program;
the processor is configured to implement the method steps of any of the above data enhancement methods when executing the program stored in the memory.
In yet another aspect of the present invention, there is also provided a computer-readable storage medium having stored therein instructions, which when run on a computer, cause the computer to perform any of the data enhancement methods described above.
In yet another aspect of the present invention, the present invention also provides a computer program product containing instructions, which when run on a computer, causes the computer to execute any of the data enhancement methods described above.
According to the data enhancement method, the data enhancement device and the electronic equipment, the oblique angle and the gray value of the image to be enhanced are adjusted to obtain oblique images under different illumination conditions, and the obtained images are enhanced images. Therefore, in the scheme, the image is adjusted from two aspects of the image inclination angle and the gray value, the two aspects of adjustment do not affect the content of the image, the availability of the obtained enhanced image is high, and the data enhancement effect is improved.
Of course, it is not necessary for any product or method of practicing the invention to achieve all of the above-described advantages at the same time.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic overall flow chart of data enhancement provided by an embodiment of the present invention;
FIG. 2 is a diagram illustrating random cropping according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of an image perspective transformation process provided by an embodiment of the invention;
FIG. 4 is a schematic diagram illustrating an image gamma correction process according to an embodiment of the present invention;
fig. 5 is a schematic diagram illustrating an effect achieved by the data enhancement method according to the embodiment of the present invention;
fig. 6 is a schematic flowchart of a data enhancement method according to an embodiment of the present invention;
FIG. 7 is a schematic flow chart of a data enhancement method according to an embodiment of the present invention;
FIG. 8 is a schematic structural diagram of a data enhancement apparatus according to an embodiment of the present invention;
fig. 9 is a schematic diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the 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.
The terms of the present invention are explained as follows:
data augmentation (Data augmentation): and expanding a sample image set of the recognition model by transforming the sample images.
Perspective Transformation (Perspective Transformation): the image is projected onto a new Viewing Plane (Viewing Plane), also called projection Mapping.
Gamma Correction (Gamma Correction): the gamma curve of the image is subjected to nonlinear adjustment, so that the contrast effect of the image is improved.
The invention concept of the invention is as follows:
the technical scheme of the invention mainly comprises three parts of random cutting, perspective transformation and gamma correction, and license plate images under different inclination angles and different illumination conditions can be obtained through the processing of the three parts and used as enhanced images to expand sample images for training license plate recognition models.
The overall data enhancement flow is shown in fig. 1, and the following describes the three parts of random cropping, perspective transformation and gamma correction in detail:
1) random cutting
And acquiring an original image, wherein the original image comprises a license plate. And calibrating the position of the license plate in the original image, and cutting the license plate from the original image according to the calibrated position. In the process of cutting, boundary disturbance is added at random to the boundary of the region where the license plate is located, and the license plate in the obtained cut image can be located at different positions of the image. As shown in fig. 2, the inner frame is a calibrated license plate region, and the outer frame is a license plate region with added boundary disturbance, that is, a clipping region.
According to the scheme, the license plate area is cut randomly to obtain various images containing license plates at different positions. The obtained images containing license plates at different positions are used for training a Convolutional Neural Network (CNN), so that the recognition capability of the CNN can be improved.
2) Perspective transformation
Through perspective transformation, license plate images with normal inclination angles are transformed into license plate images with different inclination angles, and license plates with different inclination angles which often appear at the entrance and exit of a parking lot can be simulated.
Perspective transformation, which may be understood as projecting an image onto a new Viewing Plane (Viewing Plane), is also called projection Mapping (projection Mapping). The perspective transformation formula may be:
[x′,y′,w′]＝[u,v,w]T
wherein u and v are respectively the abscissa and ordinate of the cropped image, i.e. the x-axis coordinate and the y-axis coordinate in a cartesian three-dimensional rectangular coordinate system, w is the vertical coordinate of the cropped image, i.e. the z-axis coordinate in a cartesian three-dimensional rectangular coordinate system, w is 1, (x ', y', w ') is the three-dimensional coordinate of the transformed oblique image, the three-dimensional coordinate (x', y ', w') is converted into the two-dimensional coordinate (x, y), wherein x is x '/w', y is y '/w', the matrix T is a transformation matrix,
a in matrix T33The value of (d) may be 1, and the other elements in the matrix T may be calculated as follows:
the formula: x '/w', y '/w' (equation 1), which in turn may be written as equation 2, as shown below:
and acquiring four groups of two-dimensional coordinate values in the first experimental image and four groups of two-dimensional coordinate values in the second experimental image, substituting the four groups of two-dimensional coordinate values into the formula 2 for solving, and obtaining all elements in the transformation matrix T so as to obtain the transformation matrix T, wherein the first experimental image is a cut image before transformation, and the second experimental image is an inclined image to be obtained after transformation.
And after the transformation matrix is obtained, obtaining the coordinate value of each pixel point in the image needing perspective transformation, namely the coordinate value of each pixel point in the cut image, and substituting the obtained coordinate value into a perspective transformation formula to obtain the coordinate value of the transformed inclined image.
The left inclined transformation matrix T can be obtained by calculationLAnd right tilt transformation matrix TRTo simulate a left-inclined license plate image and a right-inclined license plate image, respectively. Left oblique transformation matrix TLAnd right tilt transformation matrix TRRespectively as follows:
wherein the content of the first and second substances,
other elements may be calculated in the following manner:
transforming the matrix T according to the left slantLEquation 2 can in turn be written as:
transforming the matrix T according to the right tiltREquation 2 can in turn be written as:
obtaining four groups of two-dimensional coordinate values in the first experiment image and four groups of two-dimensional coordinate values which can form a left-inclined quadrangle in the second experiment image, substituting the four groups of two-dimensional coordinate values into the formula 3 to solve, and obtaining a left-inclined transformation matrix TLTo obtain a left tilt transformation matrix TLSimilarly, four groups of two-dimensional coordinate values in the first experimental image and four groups of two-dimensional coordinate values in the second experimental image which can form a right-leaning quadrangle are obtained and substituted into the formula 4 for solving, and then the right-leaning transformation matrix T can be obtainedRTo obtain a right tilt transformation matrix TR。
The perspective transformation formula can be written as a left oblique transformation formula as: [ x ', y ', w ']＝[u,v,w]TL
The perspective transformation formula can also be written asThe right tilt transformation formula is: [ x ', y ', w ']＝[u,v,w]TR
In the scheme, during each conversion, coordinate values of pixel points in the cut image are obtained, different angle disturbance parameters are added to the obtained coordinate values, then the coordinate values after the angle disturbance parameters are added are substituted into a left-inclination transformation formula to obtain images with different left-inclination angles, or the coordinate values after the angle disturbance parameters are added are substituted into a right-inclination transformation formula to obtain images with different right-inclination angles. The specific transformation effect is shown in fig. 3, 301 is a cropped image before transformation, different angle disturbance parameters are added to the coordinates of the cropped image, then two images with different left tilt angles after transformation 302 are obtained through a left tilt transformation formula, different angle disturbance parameters are added to the coordinates of the cropped image, and then two images with different right tilt angles after transformation 303 are obtained through a right tilt transformation formula.
3) Gamma correction
Gamma correction can achieve a contrast enhancement effect on the image to simulate different illumination intensities. The gamma correction algorithm can be understood as a power function with respect to gray scale, and the specific formula is as follows:
f(I)＝Iγ
wherein, I is the normalized input pixel gray value of the image, I ∈ [0,1], f (I) is the output pixel gray value of the image, f (I) ∈ [0,1], gamma is the image gray value adjusting coefficient, the gray value of the image can be changed by changing the size of gamma, thereby simulating different gray values of the image.
From the characteristics of the gamma curve, when γ is equal to 1, the curve is a straight line forming 45 ° with the coordinate axis, which represents that the input and output pixel gray levels are equal, when γ is less than 1, the dynamic range of the low gray level of the output pixel is increased, the dynamic range of the high gray level is decreased, the output pixel gray level of the image is increased, and the analog illumination intensity is enhanced]When f (I) is in the range of [0,0.218 ]]1/2.2 and I ∈ [0, 0.2%]When f (I) varies in the range of [0,0.5 ]]It can be seen that when γ is less than 1, the dynamic range of the low gray scale values of the output pixels increases when γ is 1 and I ∈ [0.8,1]When f (I) varies in the range of [0.8, 1]]1/2.2 and I ∈ [0.8,1 ═ y]When f (I) varies in the range of [0.9, 1]]It can be seen that when γ is less than 1, the dynamic range of the high gray value of the output pixel decreases. In this case, f (I) ═ I1/2.2＞f(I)＝I1Therefore, the output pixel gray value of the image becomes larger, i.e., the illumination intensity is enhanced. Conversely, when γ is greater than 1, the output pixel gradation value of the image becomes small, and the analog illumination intensity decreases.
In one case, random values between [0.5,2] may be used to simulate random illumination variations. When the gamma value of the gamma-corrected image is less than 1, the gray value of the output pixel is greater than that of the input pixel, and the gray value of the corrected image is greater than that of the uncorrected image, so that a scene with enhanced illumination can be simulated; when the gamma value of the gamma-corrected image is greater than 1, the gray value of the output pixel is smaller than that of the input pixel, and the gray value of the corrected image is smaller than that of the uncorrected image, so that a scene with weak illumination can be simulated. Specific correction effects as shown in fig. 4, 401 in fig. 4 is an original oblique image, 402 is a corrected image, when the original oblique image is gamma-corrected by γ less than 1, an image with a larger gray value is obtained, and when the original oblique image is gamma-corrected by γ more than 1, an image with a smaller gray value is obtained.
Random cutting, perspective transformation and gamma correction are combined, the perspective transformation can simulate images inclining to the left or the right, the gamma correction is carried out on the inclined images, the inclined images under different illumination can be simulated, and therefore the effect of data enhancement is achieved. The data enhancement effect is specifically realized as shown in fig. 5, the top image in fig. 5 is an original image, that is, an image to be enhanced, the three images in the second layer are images obtained by randomly cropping the original image to be enhanced, the image in the third layer is an oblique image obtained by performing perspective transformation on the image obtained by random cropping, and the image in the fourth layer is an image with different gray values obtained by performing gamma correction on the oblique image.
By applying the scheme, random cutting is carried out on the license plate image to be enhanced to obtain a cut image, perspective transformation is carried out on the cut image, the image inclination angle is adjusted to obtain a cut inclined image, gamma correction is carried out on the cut inclined image, the image gray value is adjusted to obtain the cut inclined image under different illumination conditions, and the cut inclined image is used as the generated enhanced image. According to the scheme, the image is adjusted from three aspects of the position, the inclination angle and the gray value of the image, the three aspects do not influence the content of the image, the obtained enhanced image is consistent with the image obtained from a real scene, the scene is various, and the usability is high, so that the data enhancement effect is improved.
In some related enhancement schemes, data enhancement is performed by means of rotation and adding noise. However, rotation and adding noise can produce samples that are severely inconsistent with the real scene. In the scheme, the image is adjusted from two aspects of the image inclination angle and the gray value, the images under different acquisition angles and different illumination conditions in the real scene are simulated, and the image with higher correlation degree with the real scene is obtained.
Based on the same inventive concept, the embodiment of the invention provides a data enhancement method, a data enhancement device and electronic equipment. The method and the device can be applied to various electronic devices, and are not limited specifically.
A data enhancement method provided in an embodiment of the present invention is described below.
Fig. 6 is a schematic flow chart of a data enhancement method according to an embodiment of the present invention, which may include:
s601: acquiring an image to be enhanced;
s602: adjusting the inclination angle of the image to be enhanced to obtain an inclined image;
s603: and adjusting the gray value of the oblique image to obtain the oblique image under different illumination conditions as a generated enhanced image.
By applying the embodiment shown in fig. 6, the oblique images with different oblique angles and different gray values are obtained as the enhanced images by adjusting the oblique angles and the gray values of the images to be enhanced. Because the image is adjusted from two aspects of the image inclination angle and the gray value, the two aspects of adjustment do not influence the content of the image, and the obtained enhanced image has high usability. Therefore, the data enhancement effect is improved.
The following is a detailed description of the steps in the embodiment of FIG. 6:
s601: and acquiring an image to be enhanced.
For example, the image to be enhanced acquired in this embodiment may be a license plate image, a face image, or an image including characters, and is not limited specifically. If the acquired image to be enhanced is a license plate image, the image which is shot by the road camera and contains the license plate can be used as the image to be enhanced. If the acquired image to be enhanced is a face image, the face image acquired by the face snapshot machine can be used as the image to be enhanced. Etc., are not to be enumerated.
S602: and adjusting the inclination angle of the image to be enhanced to obtain an inclined image.
As an embodiment, S602 may include: and performing perspective transformation on the image to be enhanced, and adjusting the inclination angle of the image to be enhanced to obtain an inclined image. For example, if the image to be enhanced is a license plate image, the license plate image can be subjected to perspective transformation to adjust the inclination angle of the license plate image, so as to obtain the license plate image with the inclination angle.
Furthermore, the image to be enhanced can be adjusted to be inclined to the left through a perspective transformation algorithm according to the left inclination transformation matrix, so that a left inclined image is obtained; and adjusting the image to be enhanced to be inclined to the right according to the right inclination transformation matrix to obtain a right inclined image. Of course, the oblique image may be obtained in other ways, such as affine transformation.
For example, the perspective transformation may be represented as:
[x′,y′,w′]＝[u,v,w]T
wherein u and v are the sit-ups of the cropped image, respectivelyThe coordinates of the scale and ordinate, i.e. the x-axis coordinate and the y-axis coordinate in a cartesian three-dimensional rectangular coordinate system, w is the vertical coordinate of the cropped image, i.e. the z-axis coordinate in a cartesian three-dimensional rectangular coordinate system, w is 1, (x ', y', w ') is the three-dimensional coordinate of the transformed tilted image, the three-dimensional coordinates (x', y ', w') are converted into two-dimensional coordinates (x, y), wherein x is x '/w', y is y '/w', the matrix T is a transformation matrix,
a in matrix T33The value of (d) may be 1, and the other elements in the matrix T may be calculated as follows:
the formula: x '/w', y '/w' (equation 1), which in turn may be written as equation 2, as shown below:
and acquiring four groups of two-dimensional coordinate values in the first experimental image and four groups of two-dimensional coordinate values in the second experimental image, substituting the four groups of two-dimensional coordinate values into the formula 2 for solving, and obtaining all elements in the transformation matrix T so as to obtain the transformation matrix T, wherein the first experimental image is a cut image before transformation, and the second experimental image is an inclined image to be obtained after transformation.
And after the transformation matrix is obtained, obtaining the coordinate value of each pixel point in the image needing perspective transformation, namely the coordinate value of each pixel point in the cut image, and substituting the obtained coordinate value into a perspective transformation formula to obtain the coordinate value of the transformed inclined image.
The left inclined transformation matrix T can be obtained by calculationLAnd right tilt transformation matrix TRTo simulate a left-inclined license plate image and a right-inclined license plate image, respectively. Left oblique transformation matrix TLAnd right tilt transformation matrix TRRespectively as follows:
wherein the content of the first and second substances,
other elements may be calculated in the following manner:
transforming the matrix T according to the left slantLEquation 2 can in turn be written as:
transforming the matrix T according to the right tiltREquation 2 can in turn be written as:
obtaining four groups of two-dimensional coordinate values in the first experiment image and four groups of two-dimensional coordinate values which can form a left-inclined quadrangle in the second experiment image, substituting the four groups of two-dimensional coordinate values into the formula 3 to solve, and obtaining a left-inclined transformation matrix TLTo obtain a left tilt transformation matrix TLSimilarly, four groups of two-dimensional coordinate values in the first experimental image and four groups of two-dimensional coordinate values in the second experimental image which can form a right-leaning quadrangle are obtained and substituted into the formula 4 for solving, and then the right-leaning transformation matrix T can be obtainedRTo obtain a right tilt transformation matrix TR。
The perspective transformation formula can be written as a left oblique transformation formula as: [ x ', y ', w ']＝[u,v,w]TL
The perspective transformation formula can also be written as a right-oblique transformation formula, which is: [ x ', y ', w ']＝[u,v,w]TR
In the scheme, during each conversion, coordinate values of all pixel points in the cut image are obtained, different angle disturbance parameters are added to the obtained coordinate values, then the coordinate values after the angle disturbance parameters are added are substituted into a left-inclination transformation formula to obtain images with different left-inclination angles, or the coordinate values after the angle disturbance parameters are added are substituted into a right-inclination transformation formula to obtain images with different right-inclination angles.
S603: and adjusting the gray value of the oblique image to obtain the oblique image under different illumination conditions as a generated enhanced image.
As an embodiment, S603 may include: and carrying out nonlinear adjustment on the gray value in the oblique image by utilizing a gamma correction algorithm to obtain the oblique image under different illumination conditions as a generated enhanced image.
Gamma (Gamma) correction is understood as a method of editing a Gamma curve of an image to perform nonlinear tone editing on the image, and a dark portion and a light portion in an image signal are detected and increased in proportion, thereby improving the image contrast effect. The gamma curve is a special tone curve.
The gamma correction algorithm is as follows:
f(I)＝Iγ
wherein, I is the normalized input pixel gray value of the image, I ∈ [0,1], f (I) is the output pixel gray value of the image, f (I) ∈ [0,1], gamma is the image gray value adjusting coefficient, the gray value of the image can be changed by changing the size of gamma, thereby simulating different gray values of the image.
From the characteristics of the gamma curve, when γ is equal to 1, the gamma curve is a straight line forming 45 ° with the coordinate axis, which represents that the input and output pixel gray values are equal, when γ is less than 1, the dynamic range of the low gray value of the output pixel is increased, the dynamic range of the high gray value is decreased, the output pixel gray value of the image is increased, and the analog illumination intensity is enhanced]When f (I) is in the range of [0,0.218 ]]1/2.2 and I ∈ [0, 0.2%]When f (I) varies in the range of [0,0.5 ]]It can be seen that when γ is less than 1, the dynamic range of the low gray scale values of the output pixels increases when γ is 1 and I ∈ [0.8,1]When f (I) varies in the range of [0.8, 1]]1/2.2 and I ∈ [0.8,1 ═ y]When f (I) varies in the range of [0.9, 1]]It can be seen that when γ is less than 1, the dynamic range of the high gray value of the output pixel decreases. In this case, f (I) ═ I1/2.2 > f (I) ═ I1Thus, the output pixel gray scale value of the image becomes large, i.e.The illumination intensity is enhanced. Conversely, when γ is greater than 1, the output pixel gradation value of the image becomes small, and the analog illumination intensity decreases.
In one case, random values between [0.5,2] may be used to simulate random illumination variations. When the gamma value of the gamma-corrected image is less than 1, the gray value of the output pixel is greater than that of the input pixel, and the gray value of the corrected image is greater than that of the uncorrected image, so that a scene with enhanced illumination can be simulated; when the gamma value of the gamma-corrected image is greater than 1, the gray value of the output pixel is smaller than that of the input pixel, and the gray value of the corrected image is smaller than that of the uncorrected image, so that a scene with weak illumination can be simulated.
Of course, the adjustment of the gray level of the image may also be performed by other manners, such as reversing the image to reverse the black and white of the image; or, a logarithmic transformation, expanding the low gray value part of the image, compressing the high gray value part of the image.
As an embodiment, after S601, an identification target area of the image to be enhanced may be determined, and the image to be enhanced is cut for multiple times with the identification target area as a reference, so as to obtain multiple cut images; then, S602 is executed to adjust the tilt angle of the image to be enhanced, so as to obtain a tilted image. In this case, in S602, specifically, the tilt angles of the plurality of cut images may be adjusted to obtain a plurality of tilt images.
By applying the embodiment, the image is cut according to the position of the recognition target in the image to obtain cut images of various scenes, the cut images are processed to generate enhanced images, the obtained enhanced images are consistent with the images acquired from the real scenes, and the scenes are various, so that the scene diversification of the enhanced images is improved, and the data enhancement effect is also improved.
Further, the identification target area of the image to be enhanced may be determined, different boundary perturbation parameters may be added to the boundary of the identification target area, respectively, to obtain different identification target frames, each identification target frame may be cut out from the image to be enhanced, as a cut image, and then S602 and S603 may be executed.
For example, if the image to be enhanced is an image containing a license plate, and the target recognition area is a license plate area, the license plate area is recognized, the recognized license plate area is calibrated by using a license plate area frame, different offset values are added to coordinate values of the license plate area frame to obtain different new license plate area frames, and the image to be enhanced is cut according to the different new license plate area frames to obtain a plurality of license plate images.
In another case, the order of S602 and S603 may be changed, that is, after the image to be enhanced is obtained, the gamma correction algorithm may be first used to perform nonlinear adjustment on the gray value of the image to be enhanced to obtain images under different illumination conditions, and then perspective transformation is performed on the images under different illumination conditions to adjust the inclination angles of the images under different illumination conditions to obtain inclined images under different illumination conditions, which are used as the generated enhanced images.
In another case, the step of determining the recognition target area of the image to be enhanced and performing multiple cropping on the image to be enhanced with reference to the recognition target area to obtain multiple cropped images may be performed after S602, or may be performed after S603.
That is, after the image to be enhanced is obtained, the inclination angle of the image to be enhanced is adjusted to obtain an inclined image, the inclined image is cut to obtain a cut inclined image, and then the gray value of the cut inclined image is adjusted to obtain cut inclined images under different illumination conditions, which are used as the generated enhanced image.
Or after the image to be enhanced is obtained, firstly carrying out nonlinear adjustment on the gray value of the image to be enhanced to obtain images under different illumination conditions, then cutting the images under different illumination conditions to obtain cut images under different illumination conditions, and then adjusting the inclination angles of the cut images under different illumination conditions to obtain cut inclined images under different illumination conditions to serve as the generated enhanced image.
When the image to be enhanced is a license plate image, obtaining oblique images under different illumination conditions by adjusting the gray value of the oblique image, and taking the oblique images as the generated enhanced image, further comprising: and adding the generated enhanced image to a sample training set of the license plate recognition model.
The method comprises the steps of training by utilizing a sample training set to obtain a license plate recognition model, and generally collecting a large number of license plate images in order to improve recognition accuracy of the recognition model. According to the scheme, the license plate images of the samples are collected, the inclination angles and the gray values of the collected license plate images of the samples are adjusted to obtain the license plate images under different specific inclination angles and different illumination conditions, the sample training set is expanded, and the cost for collecting a large number of license plate images of the samples is reduced.
Furthermore, in another flow chart of data enhancement, an embodiment of the present invention provides a process diagram of data enhancement, where in the embodiment, an image to be enhanced is first randomly cropped, and then an inclination angle and a gray value of the randomly cropped image are adjusted to obtain an enhanced image. Specifically, as shown in fig. 7, the method may include:
s701: acquiring an image to be enhanced;
the image to be enhanced acquired in this embodiment may be a license plate image, a face image, or an image including characters, and is not particularly limited.
S702: determining an identification target area in the image to be enhanced;
the recognition target area may be a license plate area, a character area, a face area, or the like.
S703: taking the identification target area as a reference, and cutting the image to be enhanced for multiple times to obtain multiple cut images;
taking the recognition target area as a reference, cutting the image to be enhanced for multiple times to obtain multiple cut images, wherein different boundary disturbance parameters can be added to the boundary of the recognition target area respectively to obtain different recognition target frames; and respectively cutting out each recognition target image from the image to be enhanced according to different recognition target frames to obtain a cut image.
For example, the position of the license plate in the original license plate image is calibrated, a boundary disturbance parameter is added to the boundary of the calibrated license plate position to obtain a new license plate recognition frame, and then the license plate image calibrated by the new license plate recognition frame is cut from the original image according to the new license plate recognition frame to serve as the cut image. Different boundary disturbance parameters are added to the boundary of the license plate position calibrated in the original license plate image, and images containing license plates at different positions can be obtained by cutting. Therefore, the effect of obtaining various scene sample images by random cutting is achieved, and the phenomenon of obtaining a single scene sample image by random cutting in the existing scheme is avoided.
S704: adjusting the inclination angle of the cut image by carrying out perspective transformation on the cut image to obtain an inclined image;
by obtaining coordinate values of pixel points in the cut image, different angle disturbance parameters are added to the coordinates, and then the coordinate values added with the different angle disturbance parameters can be substituted into a left inclination transformation formula and a right inclination transformation formula to obtain images with different inclination angles.
S705: and carrying out nonlinear adjustment on the gray value in the oblique image by utilizing a gamma correction algorithm to obtain the oblique image under different illumination conditions as a generated enhanced image.
Acquiring a gray value of an inclined image, performing normalization processing on the gray value, taking the acquired normalized pixel gray value as an input pixel gray value of a gamma correction algorithm, obtaining different output pixel gray values by changing the gamma value in the gamma correction algorithm, wherein the range of the gamma value can be [0.5,2], and taking the images under different illumination conditions as generated enhanced images corresponding to the images with different gray values, namely the images under different illumination conditions.
If the generated enhanced image is generated by a license plate image, and the recognition model is a license plate recognition model, the enhanced images with different inclination angles and different gray values generated by the license plate image can be added into a sample license plate image training set of the license plate recognition model after the enhanced image is generated.
By applying the embodiment shown in fig. 7, the image to be enhanced is cut randomly to obtain a cut image, then the cut image is subjected to perspective transformation, the image inclination angle is adjusted to obtain a cut inclined image, then the cut inclined image is subjected to gamma correction, the image gray value is adjusted to obtain cut inclined images under different illumination conditions, and the cut inclined images are used as the generated enhanced image. Because the image is adjusted from the three aspects of the position, the inclination angle and the gray value of the image, and the three aspects do not influence the content of the image, the obtained enhanced image is consistent with the image obtained from the real scene, the scene is various, and the usability is high. Therefore, the data enhancement effect is improved, and the phenomenon that sample images which are seriously inconsistent with real scenes are generated due to single scene of the sample images obtained in the conventional scheme is avoided.
Corresponding to the method embodiment shown in fig. 6, an embodiment of the present invention further provides a data enhancement apparatus, as shown in fig. 8, where the apparatus includes:
an image obtaining module 801, configured to obtain an image to be enhanced;
an angle adjusting module 802, configured to adjust an inclination angle of the image to be enhanced to obtain an inclined image;
a gray scale adjusting module 803, configured to obtain the oblique image under different illumination conditions as the generated enhanced image by adjusting the gray scale value of the oblique image.
As an implementation manner, the angle adjusting module 802 may be specifically configured to perform perspective transformation on the image to be enhanced, and adjust an inclination angle of the image to be enhanced, so as to obtain an inclined image.
As an embodiment, the angle adjustment module 802 may include: a left tilt transform submodule (not shown) and a right tilt transform submodule (not shown);
a left-oblique transformation submodule (not shown in the figure) for adjusting the image to be enhanced to be left-oblique according to the left-oblique transformation matrix to obtain a left-oblique image;
and a right-oblique transformation submodule (not shown in the figure) for adjusting the image to be enhanced to be oblique to the right according to the right-oblique transformation matrix to obtain a right-oblique image.
As an embodiment, the gray scale adjustment module 803 may be specifically configured to perform a non-linear adjustment on the gray scale value in the oblique image by using a gamma correction algorithm, so as to obtain the oblique image under different illumination conditions as the generated enhanced image.
As an embodiment, the apparatus may further include: a region determining module (not shown in the figure) and a cutting module (not shown in the figure);
a region determining module (not shown in the figure) configured to determine a recognition target region in the image to be enhanced after the acquiring of the image to be enhanced;
a cropping module (not shown in the figure) for cropping the image to be enhanced for multiple times by taking the recognition target area as a reference to obtain multiple cropped images;
the angle adjusting module 802 is specifically configured to adjust the tilt angles of the multiple cut images respectively to obtain multiple tilted images.
As an embodiment, the cutting module (not shown in the figure) may include: a boundary perturbation adding submodule (not shown in the figure) and a clipping submodule (not shown in the figure);
a boundary disturbance adding submodule (not shown in the figure) for adding different boundary disturbance parameters to the boundary of the recognition target area to obtain different recognition target frames;
and a cropping submodule (not shown in the figure) for cropping each recognition target frame from the image to be enhanced respectively as a cropped image.
As an embodiment, the image to be enhanced is a license plate image, and the apparatus may further include: an image adding module (not shown in the figure);
and an image adding module (not shown in the figure) for adding the generated enhanced image to a sample training set of the license plate recognition model after the oblique image under different illumination conditions is obtained by adjusting the gray value of the oblique image and is used as the generated enhanced image.
By applying the embodiment shown in fig. 8, the oblique images with different oblique angles and different gray values are obtained as the enhanced images by adjusting the oblique angles and the gray values of the images to be enhanced. Because the image is adjusted from two aspects of the image inclination angle and the gray value, the two aspects of adjustment do not influence the content of the image, and the obtained enhanced image has high usability. Therefore, the data enhancement effect is improved.
An embodiment of the present invention further provides an electronic device, as shown in fig. 9, which includes a processor 901, a communication interface 902, a memory 903, and a communication bus 904, where the processor 901, the communication interface 902, and the memory 903 complete mutual communication through the communication bus 904,
a memory 903 for storing computer programs;
the processor 901 is configured to implement the following steps when executing the program stored in the memory 903:
acquiring an image to be enhanced;
adjusting the inclination angle of the image to be enhanced to obtain an inclined image;
and adjusting the gray value of the oblique image to obtain the oblique image under different illumination conditions as a generated enhanced image.
Therefore, in the scheme provided by the embodiment of the invention, the inclined images with different inclined angles and different gray values are obtained as the enhanced images by adjusting the inclined angles and the gray values of the images to be enhanced. Because the image is adjusted from two aspects of the image inclination angle and the gray value, the two aspects of adjustment do not influence the content of the image, and the obtained enhanced image has high usability. Therefore, the data enhancement effect is improved.
The communication bus mentioned in the electronic device may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown, but this does not mean that there is only one bus or one type of bus.
The communication interface is used for communication between the electronic equipment and other equipment.
The Memory may include a Random Access Memory (RAM) or a Non-Volatile Memory (NVM), such as at least one disk Memory. Optionally, the memory may also be at least one memory device located remotely from the processor.
The Processor may be a general-purpose Processor, including a Central Processing Unit (CPU), a Network Processor (NP), and the like; but may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, discrete hardware component.
In yet another embodiment of the present invention, a computer-readable storage medium is further provided, which has instructions stored therein, which when run on a computer, cause the computer to perform the data enhancement method described in any of the above embodiments.
In a further embodiment of the present invention, there is also provided a computer program product containing instructions which, when run on a computer, cause the computer to perform the data enhancement method of any of the above embodiments.
In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, cause the processes or functions described in accordance with the embodiments of the invention to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, from one website site, computer, server, or data center to another website site, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that incorporates one or more of the available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., Solid State Disk (SSD)), among others.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
All the embodiments in the present specification are described in a related manner, and the same and similar parts among the embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, as for the apparatus embodiment, the device embodiment and the computer storage medium embodiment, since they are substantially similar to the method embodiment, the description is relatively simple, and the relevant points can be referred to the partial description of the method embodiment.
The above description is only for the preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention shall fall within the protection scope of the present invention.
Claims (16)
1. A method of data enhancement, the method comprising:
acquiring an image to be enhanced;
adjusting the inclination angle of the image to be enhanced to obtain an inclined image;
and adjusting the gray value of the oblique image to obtain the oblique image under different illumination conditions as a generated enhanced image.
2. The method according to claim 1, wherein the adjusting the tilt angle of the image to be enhanced to obtain a tilted image comprises:
and adjusting the inclination angle of the image to be enhanced by carrying out perspective transformation on the image to be enhanced to obtain an inclined image.
3. The method according to claim 2, wherein the adjusting the tilt angle of the image to be enhanced by performing perspective transformation on the image to be enhanced to obtain a tilted image comprises:
according to the left inclination transformation matrix, adjusting the image to be enhanced to incline to the left to obtain a left inclined image;
and adjusting the image to be enhanced to be inclined to the right according to the right inclination transformation matrix to obtain a right inclined image.
4. The method according to claim 1, wherein the obtaining of the oblique image under different illumination conditions by adjusting the gray-scale value of the oblique image as the generated enhanced image comprises:
and carrying out nonlinear adjustment on the gray value in the oblique image by utilizing a gamma correction algorithm to obtain the oblique image under different illumination conditions as a generated enhanced image.
5. The method according to claim 1, further comprising, after said acquiring an image to be enhanced:
determining an identification target area in the image to be enhanced;
taking the identification target area as a reference, and cutting the image to be enhanced for multiple times to obtain multiple cut images;
the adjusting the inclination angle of the image to be enhanced to obtain an inclined image comprises:
and adjusting the inclination angles of the plurality of cut images respectively to obtain a plurality of inclined images.
6. The method according to claim 5, wherein the cropping the image to be enhanced for a plurality of times with reference to the recognition target area to obtain a plurality of cropped images comprises:
adding different boundary disturbance parameters to the boundary of the identification target area respectively to obtain different identification target frames;
and respectively cutting out each recognition target frame from the image to be enhanced to be used as a cut image.
7. The method according to claim 1, wherein the image to be enhanced is a license plate image;
after the obtaining of the oblique images under different illumination conditions by adjusting the gray-scale values of the oblique images as the generated enhanced images, the method further includes:
and adding the generated enhanced image to a sample training set of a license plate recognition model.
8. A data enhancement apparatus, characterized in that the apparatus comprises:
the image acquisition module is used for acquiring an image to be enhanced;
the angle adjusting module is used for adjusting the inclination angle of the image to be enhanced to obtain an inclined image;
and the gray level adjusting module is used for adjusting the gray level value of the oblique image to obtain the oblique image under different illumination conditions as the generated enhanced image.
9. The apparatus according to claim 8, wherein the angle adjusting module is specifically configured to adjust an inclination angle of the image to be enhanced by performing perspective transformation on the image to be enhanced, so as to obtain an inclined image.
10. The apparatus of claim 9, wherein the angle adjustment module comprises:
the left inclination transformation submodule is used for adjusting the image to be enhanced to incline to the left according to a left inclination transformation matrix to obtain a left inclination image;
and the right inclination transformation submodule is used for adjusting the image to be enhanced to be inclined to the right according to the right inclination transformation matrix to obtain a right inclined image.
11. The apparatus according to claim 8, wherein the gray scale adjustment module is specifically configured to perform a non-linear adjustment on the gray scale value in the oblique image by using a gamma correction algorithm, so as to obtain the oblique image under different illumination conditions as the generated enhanced image.
12. The apparatus of claim 8, further comprising:
the region determining module is used for determining a recognition target region in the image to be enhanced after the image to be enhanced is obtained;
the cutting module is used for cutting the image to be enhanced for multiple times by taking the identification target area as a reference to obtain multiple cut images;
the angle adjusting module is specifically configured to adjust the tilt angles of the plurality of cut images respectively to obtain a plurality of tilted images.
13. The apparatus of claim 12, wherein the cropping module comprises:
the boundary disturbance adding submodule is used for respectively adding different boundary disturbance parameters to the boundary of the identification target area to obtain different identification target frames;
and the cutting submodule is used for respectively cutting each recognition target frame from the image to be enhanced to be used as a cut image.
14. The apparatus of claim 8, wherein the image to be enhanced is a license plate image, and the apparatus further comprises:
and the image adding module is used for adjusting the gray value of the oblique image to obtain the oblique image under different illumination conditions, and adding the generated enhanced image to a sample training set of the license plate recognition model after the oblique image is used as the generated enhanced image.
15. An electronic device, characterized in that the device comprises: the system comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory are communicated with each other through the communication bus;
the memory is used for storing a computer program;
the processor is configured to implement the method steps of any one of claims 1 to 7 when executing the program stored in the memory.
16. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the method steps of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811526923.5A CN111325670A (en) | 2018-12-13 | 2018-12-13 | Data enhancement method and device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811526923.5A CN111325670A (en) | 2018-12-13 | 2018-12-13 | Data enhancement method and device and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111325670A true CN111325670A (en) | 2020-06-23 |
Family
ID=71172319
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811526923.5A Pending CN111325670A (en) | 2018-12-13 | 2018-12-13 | Data enhancement method and device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111325670A (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5271064A (en) * | 1991-06-14 | 1993-12-14 | University Of Cincinnati | Apparatus and method for smoothing regions and enhancing edges in gray scale images |
| US20100054595A1 (en) * | 2008-08-26 | 2010-03-04 | Microsoft Corporation | Automatic Image Straightening |
| US20170083762A1 (en) * | 2015-06-22 | 2017-03-23 | Photomyne Ltd. | System and Method for Detecting Objects in an Image |
| CN106599799A (en) * | 2016-11-24 | 2017-04-26 | 厦门中控生物识别信息技术有限公司 | Sample generation method and device for face detection |
| CN107527333A (en) * | 2017-07-31 | 2017-12-29 | 湖北工业大学 | A kind of rapid image Enhancement Method based on gamma transformation |
| CN107679525A (en) * | 2017-11-01 | 2018-02-09 | 腾讯科技（深圳）有限公司 | Image classification method, device and computer-readable recording medium |
| CN108230233A (en) * | 2017-05-16 | 2018-06-29 | 北京市商汤科技开发有限公司 | Data enhancing, treating method and apparatus, electronic equipment and computer storage media |
| CN108597003A (en) * | 2018-04-20 | 2018-09-28 | 腾讯科技（深圳）有限公司 | A kind of article cover generation method, device, processing server and storage medium |
| CN108985137A (en) * | 2017-06-02 | 2018-12-11 | 杭州海康威视数字技术股份有限公司 | A kind of licence plate recognition method, apparatus and system |
-
2018
- 2018-12-13 CN CN201811526923.5A patent/CN111325670A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5271064A (en) * | 1991-06-14 | 1993-12-14 | University Of Cincinnati | Apparatus and method for smoothing regions and enhancing edges in gray scale images |
| US20100054595A1 (en) * | 2008-08-26 | 2010-03-04 | Microsoft Corporation | Automatic Image Straightening |
| US20170083762A1 (en) * | 2015-06-22 | 2017-03-23 | Photomyne Ltd. | System and Method for Detecting Objects in an Image |
| CN106599799A (en) * | 2016-11-24 | 2017-04-26 | 厦门中控生物识别信息技术有限公司 | Sample generation method and device for face detection |
| CN108230233A (en) * | 2017-05-16 | 2018-06-29 | 北京市商汤科技开发有限公司 | Data enhancing, treating method and apparatus, electronic equipment and computer storage media |
| CN108985137A (en) * | 2017-06-02 | 2018-12-11 | 杭州海康威视数字技术股份有限公司 | A kind of licence plate recognition method, apparatus and system |
| CN107527333A (en) * | 2017-07-31 | 2017-12-29 | 湖北工业大学 | A kind of rapid image Enhancement Method based on gamma transformation |
| CN107679525A (en) * | 2017-11-01 | 2018-02-09 | 腾讯科技（深圳）有限公司 | Image classification method, device and computer-readable recording medium |
| CN108597003A (en) * | 2018-04-20 | 2018-09-28 | 腾讯科技（深圳）有限公司 | A kind of article cover generation method, device, processing server and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 穆晓芳等: "数字图像处理技术", 煤炭工业出版社, pages: 45 - 48 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110705583B (en) | Cell detection model training method, device, computer equipment and storage medium | |
| CN109583483B (en) | Target detection method and system based on convolutional neural network | |
| CN109753971B (en) | Correction method and device for distorted text lines, character recognition method and device | |
| CN111833237B (en) | Image registration method based on convolutional neural network and local homography transformation | |
| CN110378837B (en) | Target detection method and device based on fish-eye camera and storage medium | |
| CN107172418A (en) | A kind of tone scale map image quality evaluating method analyzed based on exposure status | |
| CN110913205B (en) | Video special effect verification method and device | |
| CN109871845B (en) | Certificate image extraction method and terminal equipment | |
| CN111340077B (en) | Attention mechanism-based disparity map acquisition method and device | |
| US11875486B2 (en) | Image brightness statistical method and imaging device | |
| CN110648284B (en) | Image processing method and device with uneven illumination | |
| CN114155285B (en) | Image registration method based on gray histogram | |
| CN115082450A (en) | Pavement crack detection method and system based on deep learning network | |
| CN113516697B (en) | Image registration method, device, electronic equipment and computer readable storage medium | |
| CN111353325A (en) | Key point detection model training method and device | |
| CN111539975A (en) | Method, device and equipment for detecting moving target and storage medium | |
| US20220222813A1 (en) | Method for determining plant growth curve and electronic device | |
| CN116597246A (en) | Model training method, target detection method, electronic device and storage medium | |
| TWI803243B (en) | Method for expanding images, computer device and storage medium | |
| CN110852958A (en) | Self-adaptive correction method and device based on object inclination angle | |
| CN111325670A (en) | Data enhancement method and device and electronic equipment | |
| CN113315995B (en) | Method and device for improving video quality, readable storage medium and electronic equipment | |
| CN112149745B (en) | Method, device, equipment and storage medium for determining difficult example sample | |
| CN114742774A (en) | No-reference image quality evaluation method and system fusing local and global features | |
| CN114549932A (en) | Data enhancement processing method and device, computer equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |