CN112200258A - Multi-dimensional vehicle logo data enhancement method - Google Patents
Multi-dimensional vehicle logo data enhancement method Download PDFInfo
- Publication number
- CN112200258A CN112200258A CN202011114773.4A CN202011114773A CN112200258A CN 112200258 A CN112200258 A CN 112200258A CN 202011114773 A CN202011114773 A CN 202011114773A CN 112200258 A CN112200258 A CN 112200258A
- Authority
- CN
- China
- Prior art keywords
- image
- dimension
- original image
- car logo
- width
- 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
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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/255—Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention relates to a multi-dimensional car logo data enhancement method, which comprises the steps of firstly generating a small rectangular frame with four vertexes having offsets of 1.5 to 3 times of the width and the height of a car logo area relative to the vertexes of a car logo through a small frame random selection method, selecting an image containing the car logo area by using the rectangular frame, and intercepting the image as a scale dimension enhanced image. And then carrying out quantitative dimension enhancement, and scanning and intercepting the image by using a rectangular frame with the size of the source image 1/2 at the step length of 1/4 by a sliding cross segmentation method to generate a large number of effective quantitative dimension enhanced images. And generating a rectangular frame with the length from the center of the car logo area to the minimum value from the left side to the right side of the image and the width from the center of the car logo area to the minimum value from the upper side to the lower side of the image by a normal distribution segmentation method, selecting a part of image containing the car logo area by using the rectangular frame according to the probability of normal distribution, and intercepting the part of image to be used as a space dimensionality enhanced image. And finally, enhancing the brightness dimension and the contrast dimension of all the data by using brightness transformation and contrast transformation. The invention can effectively expand the data volume of the car logo.
Description
Technical Field
The invention relates to the field of data enhancement of pattern recognition and computer vision, in particular to a multi-dimensional car logo data enhancement method.
Background
In the deep learning field, big data is the basis for supporting object feature learning. Training the network requires a large amount of data as support in order to better extract the features of the target. If the data quality in the data set is not good enough, it is usually optimized using methods such as data equalization or data enhancement. The significance of data enhancement is that training data are converted through a certain method to generate new data, so that the network can learn features better. Through data enhancement, the original data set can be enriched and the data set can be expanded. It can prevent overfitting due to small amount of data during training. This has important significance to the recognition and detection capability of the model. Most of the data of the current car logo data base set are images which are separated from scenes and used for classification, and the data set cannot be used for detection training. The data set that can be used for detection training is too small to meet the large amount of data required for training. Therefore, data enhancement of the emblem data set is required to meet the demand.
Disclosure of Invention
In view of this, the present invention provides a method for enhancing data of a multi-dimensional emblem, which can effectively expand the data amount of the emblem.
The invention is realized by adopting the following scheme: a multi-dimensional car logo data enhancement method comprises the following steps:
step S1: generating a small rectangular frame with four vertexes having offsets of 1.5 to 3 times of the width and the height of the car logo region relative to the car logo vertexes by using a small frame random selection method, and generating a scale dimension enhanced image by using a method of selecting the car logo region image by using the small rectangular frame;
step S2: scanning an original image by using a cross sliding window method, wherein the length and the width of a window are 1/2 of the original image, the sliding step length is 1/4 of the original image, and then intercepting a window image containing a complete car logo area in each scanning as a number dimension enhanced image;
step S3: generating a rectangle with the length being the minimum value of the distance from the center of the car logo area to the left side and the right side of the image and the width being the minimum value of the distance from the center of the car logo area to the upper side and the lower side of the image by using a normal distribution segmentation method, selecting a window image containing the complete car logo area according to the probability of normal distribution, and intercepting the window image as a space dimension enhanced image;
step S4: the enhanced images generated in the integration step S1, step S2 and step S3 are further enhanced by using a luminance transformation method and a contrast transformation method, respectively.
Further, the step S1 specifically includes the following steps:
step S11: generating the offsets dx1, dy1, dx2, dy2 from the four sides of the rectangular frame to the four sides of the emblem area
Wherein dx1, dy1, dx2, dy2 are offsets of the left, upper, right and lower sides of the rectangular frame from the left, upper, right and lower sides of the emblem region, rw and rh are the width and height of the emblem region, respectively, random (1.5, 3) is a function that randomly generates uniformly distributed random floating point numbers between 1.5 and 3;
step S12: and calculating the position of the generated new image in the original image by the following method:
wherein sf _ xx1 and sf _ yy1 are positions of the upper left corner of the dimension-enhanced image in the original image; sf _ xx2 and sf _ yy2 are positions of the dimension enhanced image at the lower right corner of the original image, and dx1, dy1, dx2 and dy2 are offsets of the left side, the upper side, the right side and the lower side of the rectangular frame from the left side, the upper side, the right side and the lower side of the logo area respectively;
step S13, calculating the new coordinate of the center of the car logo:
wherein cx is an x coordinate of the center of the new logo, cy is a y coordinate of the center of the new logo, and sf _ xx1 and sf _ yy1 are positions of the upper left corner of the dimension-enhanced image in the original image;
step S14: generating a new car logo image by using the position of the scale dimension enhanced image calculated in the step S12, and writing the calculated car logo information into a file named by adding an sf suffix to the file name of the original image;
further, the step S2 specifically includes the following steps:
step S21: calculating the length and width of the subimages
Wherein nw and nh are respectively the width and height of the enhanced image in the number dimension, and ow and oh are respectively the width and height of the original image;
step S22: and calculating the positions of the generated number dimension enhanced images in the original image, wherein the calculation method comprises the following steps:
wherein i is 0,1, 2, j is 0,1, 2, ss _ xx1 and ss _ yy1 are positions of the upper left corner of the enhanced image in the number dimension in the original image, ss _ xx2 and ss _ yy2 are positions of the enhanced image in the lower right corner of the original image in the number dimension, and nw and nh are the width and the height of the enhanced image in the number dimension respectively;
step S23: judging whether the generated quantity dimension enhanced image contains a complete car logo area, if so, retaining the sub-image, otherwise, discarding;
step S24: a new emblem image is generated using the number dimension enhanced image positions calculated in step S22, and the calculated emblem position information is written into a file named by the original image file name plus the ss suffix.
Further, the step S3 specifically includes the following steps:
step S31: a function randn (0,1) is constructed to generate a number between 0 and 1 that should satisfy the normal distribution:
it is calculated that when μ is 0.5 and σ is 0.1938, the random number generated using the normal distribution can satisfy 0 to 1; the norm function in Matlab is used as a vector, so the randn (0,1) function is specifically constructed by using the norm function in Matlab (0.5, 0.1938);
step S32: calculating the width and height of the spatial dimension enhanced image:
wherein nw and nh are the width and height, respectively, of the spatial dimension enhanced image; ox and oy are the x coordinate and y coordinate of the center of the target area of the original image respectively; ow and oh are the width and height of the original image, respectively;
step S33: calculating the position of the central point of the car logo area in the generated space dimension enhanced image, wherein the calculation method comprises the following steps:
wherein nx and ny are respectively an x coordinate and a y coordinate of a target center point of the spatial dimension enhanced image; ow and oh are the width and height of the original image, respectively; nw and nh are the width and height, respectively, of the spatial dimension enhanced image; rw and rh are the width and height of the target region, respectively; randn (0,1) is a function that generates a number between 0 and 1 that satisfies a normal distribution;
step S34: and calculating the position of the generated space dimension enhanced image in the original image, wherein the calculation method comprises the following steps:
wherein mr _ xx1 and mr _ yy1 are positions of the upper left corner of the spatial dimension enhanced image in the original image; mr _ xx2 and mr _ yy2 are positions of the enhanced image in the lower right corner of the original image in the spatial dimension; ox and oy are the x coordinate and y coordinate of the center of the target area of the original image respectively; nx and ny are the x coordinate and the y coordinate of the target center point of the spatial dimension enhanced image, respectively; nw and nh are the width and height, respectively, of the spatial dimension enhanced image;
step S35: a new car logo image is generated using the spatial dimension enhanced image position calculated in step S34, and the calculated car logo position information is written in a file named by adding an mr suffix to the original image file name.
Further, the step S4 specifically includes the following steps:
step S41: integrating all the images generated in step S1, step S2, and step S3, copying the original image so that the ratio of the number of generated images to the number of original images is 3: 1;
step S42: performing brightness transformation processing on all the images generated in the steps S1, S2 and S3, wherein the brightness of each picture is randomly selected from 0.5 to 1.5 for transformation so as to strengthen the distribution of the data set in the brightness dimension;
step S43: performing contrast transformation processing on all the images generated in the steps S1, S2 and S3, wherein the contrast of each image is randomly selected from 0.5 to 1.5 for transformation, so as to strengthen the distribution of the data set in the contrast dimension;
step S44: all images are integrated to form a new data set.
Compared with the prior art, the invention has the following beneficial effects:
(1) the invention can effectively perform data expansion on the car logo image.
(2) The small frame random selection method provided by the invention can effectively increase the proportion of the car logo in the overall image, so that the characteristics of the car logo in the image are more prominent.
(3) The cross sliding segmentation method provided by the invention can effectively increase the number of pictures without changing the length-width ratio of the original image.
(4) The normal distribution segmentation method provided by the invention can maximally keep the information in the image on the premise of greatly increasing the diversity of the positions of the car logos.
Drawings
FIG. 1 is a flow chart of a method according to an embodiment of the present invention.
Detailed Description
The invention is further explained below with reference to the drawings and the embodiments.
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
As shown in fig. 1, the present embodiment provides a method for enhancing multidimensional car logo data, including the following steps:
step S1: generating a small rectangular frame with four vertexes having offsets of 1.5 to 3 times of the width and the height of the vehicle logo region relative to the vehicle logo vertexes by using a small frame random selection method, and generating a scale dimension enhanced image by using a method of selecting a vehicle logo region image by using the small rectangular frame, wherein the generated image can effectively increase the proportion of the vehicle logo region in the whole image from the size dimension, and highlights the characteristic information of the vehicle logo in the image;
step S2: scanning an original image by using a cross sliding window method, wherein the length and the width of a window are 1/2 of the original image, the sliding step length is 1/4 of the original image, and then intercepting a window image containing a complete car logo area in each scanning as a number dimension enhanced image;
step S3: generating a rectangle with the length being the minimum value of the distance from the center of the car logo area to the left side and the right side of the image and the width being the minimum value of the distance from the center of the car logo area to the upper side and the lower side of the image by using a normal distribution segmentation method, selecting a window image containing the complete car logo area according to the probability of normal distribution, and intercepting the window image as a space dimension enhanced image;
step S4: the enhanced images generated in the integration step S1, step S2 and step S3 are further enhanced by using a luminance transformation method and a contrast transformation method, respectively.
In this embodiment, the step S1 specifically includes the following steps:
step S11: generating the offsets dx1, dy1, dx2, dy2 from the four sides of the rectangular frame to the four sides of the emblem area
Wherein dx1, dy1, dx2, dy2 are offsets of the left, upper, right and lower sides of the rectangular frame from the left, upper, right and lower sides of the emblem region, rw and rh are the width and height of the emblem region, respectively, random (1.5, 3) is a function that randomly generates uniformly distributed random floating point numbers between 1.5 and 3;
step S12: and calculating the position of the generated new image in the original image by the following method:
wherein sf _ xx1 and sf _ yy1 are positions of the upper left corner of the dimension-enhanced image in the original image; sf _ xx2 and sf _ yy2 are positions of the dimension enhanced image at the lower right corner of the original image, and dx1, dy1, dx2 and dy2 are offsets of the left side, the upper side, the right side and the lower side of the rectangular frame from the left side, the upper side, the right side and the lower side of the logo area respectively;
step S13, calculating the new coordinate of the center of the car logo:
wherein cx is an x coordinate of the center of the new logo, cy is a y coordinate of the center of the new logo, and sf _ xx1 and sf _ yy1 are positions of the upper left corner of the dimension-enhanced image in the original image;
step S14: generating a new car logo image by using the position of the scale dimension enhanced image calculated in the step S12, and writing the calculated car logo information into a file named by adding a suffix of 'sf' to the file name of the original image;
in this embodiment, the step S2 specifically includes the following steps:
step S21: calculating the length and width of the subimages
Wherein nw and nh are respectively the width and height of the enhanced image in the number dimension, and ow and oh are respectively the width and height of the original image;
step S22: and calculating the positions of the generated number dimension enhanced images in the original image, wherein the calculation method comprises the following steps:
wherein i is 0,1, 2, j is 0,1, 2, ss _ xx1 and ss _ yy1 are positions of the upper left corner of the enhanced image in the number dimension in the original image, ss _ xx2 and ss _ yy2 are positions of the enhanced image in the lower right corner of the original image in the number dimension, and nw and nh are the width and the height of the enhanced image in the number dimension respectively;
step S23: judging whether the generated quantity dimension enhanced image contains a complete car logo area, if so, retaining the sub-image, otherwise, discarding;
step S24: a new emblem image is generated using the number-dimension enhanced image positions calculated in step S22, and the calculated emblem position information is written in a file named after the original image file name plus the "ss" suffix.
In this embodiment, the step S3 specifically includes the following steps:
step S31: a function randn (0,1) is constructed to generate a number between 0 and 1 that should satisfy the normal distribution:
it can be found by calculation that the random number generated using the normal distribution can satisfy 0 to 1 when μ is 0.5 and σ is 0.1938. The invention uses the norm function in Matlab as a carrier, so the randn (0,1) function is constructed by using the norm function in Matlab (0.5, 0.1938).
Step S32: calculating the width and height of the spatial dimension enhanced image:
wherein nw and nh are the width and height, respectively, of the spatial dimension enhanced image; ox and oy are the x coordinate and y coordinate of the center of the target area of the original image respectively; ow and oh are the width and height of the original image, respectively;
step S33: calculating the position of the central point of the car logo area in the generated space dimension enhanced image, wherein the calculation method comprises the following steps:
wherein nx and ny are respectively an x coordinate and a y coordinate of a target center point of the spatial dimension enhanced image; ow and oh are the width and height of the original image, respectively; nw and nh are the width and height, respectively, of the spatial dimension enhanced image; rw and rh are the width and height of the target region, respectively; randn (0,1) is a function that generates a number between 0 and 1 that satisfies a normal distribution;
step S34: and calculating the position of the generated space dimension enhanced image in the original image, wherein the calculation method comprises the following steps:
wherein mr _ xx1 and mr _ yy1 are positions of the upper left corner of the spatial dimension enhanced image in the original image; mr _ xx2 and mr _ yy2 are positions of the enhanced image in the lower right corner of the original image in the spatial dimension; ox and oy are the x coordinate and y coordinate of the center of the target area of the original image respectively; nx and ny are the x coordinate and the y coordinate of the target center point of the spatial dimension enhanced image, respectively; nw and nh are the width and height, respectively, of the spatial dimension enhanced image;
step S35: a new car logo image is generated using the spatial dimension enhanced image position calculated in step S34, and the calculated car logo position information is written in a file named by adding the suffix "mr" to the original image file name.
In this embodiment, the step S4 specifically includes the following steps:
step S41: integrating all the images generated in step S1, step S2, and step S3, copying the original image so that the ratio of the number of generated images to the number of original images is 3: 1;
step S42: performing brightness transformation processing on all the images generated in the steps S1, S2 and S3, wherein the brightness of each picture is randomly selected from 0.5 to 1.5 for transformation so as to strengthen the distribution of the data set in the brightness dimension;
step S43: performing contrast transformation processing on all the images generated in the steps S1, S2 and S3, wherein the contrast of each image is randomly selected from 0.5 to 1.5 for transformation, so as to strengthen the distribution of the data set in the contrast dimension;
step S44: all images are integrated to form a new data set.
Preferably, the data enhancement can be effectively carried out on the car logo data set by the embodiment. Firstly, a small-frame random selection method is used, and the method is mainly used for solving the problem that the car logo occupies a small area in the main body image. Compared with the original image, the proportion of the car logo area can be averagely improved by 3 times by using the image generated by the method of the embodiment. And then, enhancing the image by using a cross sliding segmentation method. The aspect ratio of the image generated by the method of the embodiment is consistent with that of the original image, so that the generated image is more consistent with the original data distribution, and meanwhile, the number of effective data can be greatly increased. Normal distribution segmentation was then used to enrich the positional diversity of the data. The distribution of the positions of the car logos of the images generated by the method of the embodiment meets the normal distribution, and the distribution has important significance in probability. And finally, performing data integration enhancement, and copying original data to ensure that the quantity ratio of the generated data to the original data is 3: 1, simultaneously, brightness and contrast change is carried out on all data generated by the method, and the diversity of the data is further expanded.
The above description is only a preferred embodiment of the present invention, and all equivalent changes and modifications made in accordance with the claims of the present invention should be covered by the present invention.
Claims (5)
1. A multidimensional car logo data enhancement method is characterized by comprising the following steps: the method comprises the following steps:
step S1: generating a small rectangular frame with four vertexes having offsets of 1.5 to 3 times of the width and the height of the car logo region relative to the car logo vertexes by using a small frame random selection method, and generating a scale dimension enhanced image by using a method for selecting the car logo region image by using the small rectangular frame;
step S2: scanning an original image by using a cross sliding window method, wherein the length and the width of a window are 1/2 of the original image, the sliding step length is 1/4 of the original image, and then intercepting a window image containing a complete car logo area in each scanning as a number dimension enhanced image;
step S3: generating a rectangle with the length being the minimum value of the distance from the center of the car logo area to the left side and the right side of the image and the width being the minimum value of the distance from the center of the car logo area to the upper side and the lower side of the image by using a normal distribution segmentation method, selecting a window image containing the complete car logo area according to the probability of normal distribution, and intercepting the window image as a space dimension enhanced image;
step S4: the enhanced images generated in the integration step S1, step S2 and step S3 are further enhanced by using a luminance transformation method and a contrast transformation method, respectively.
2. The multi-dimensional emblem data enhancement method of claim 1, characterized in that: the step S1 specifically includes the following steps:
step S11: generating the offsets dx1, dy1, dx2, dy2 from the four sides of the rectangular frame to the four sides of the emblem area
Wherein dx1, dy1, dx2, dy2 are offsets of the left, upper, right and lower sides of the rectangular frame from the left, upper, right and lower sides of the emblem region, rw and rh are the width and height of the emblem region, respectively, random (1.5, 3) is a function that randomly generates uniformly distributed random floating point numbers between 1.5 and 3;
step S12: and calculating the position of the generated new image in the original image by the following method:
wherein sf _ xx1 and sf _ yy1 are positions of the upper left corner of the dimension-enhanced image in the original image; sf _ xx2 and sf _ yy2 are positions of the dimension enhanced image at the lower right corner of the original image, and dx1, dy1, dx2 and dy2 are offsets of the left side, the upper side, the right side and the lower side of the rectangular frame from the left side, the upper side, the right side and the lower side of the logo area respectively;
step S13: calculating new car logo center coordinates:
wherein cx is an x coordinate of the center of the new logo, cy is a y coordinate of the center of the new logo, and sf _ xx1 and sf _ yy1 are positions of the upper left corner of the dimension-enhanced image in the original image;
step S14: a new car logo image is generated using the position of the enhanced image in the scale dimension calculated in step S12, and the calculated car logo information is written in a file named by the original image file name plus the sf suffix.
3. The multi-dimensional emblem data enhancement method of claim 1, characterized in that: the step S2 specifically includes the following steps:
step S21: calculating the length and width of the subimages
Wherein nw and nh are respectively the width and height of the enhanced image in the number dimension, and ow and oh are respectively the width and height of the original image;
step S22: and calculating the positions of the generated number dimension enhanced images in the original image, wherein the calculation method comprises the following steps:
wherein i is 0,1, 2, j is 0,1, 2, ss _ xx1 and ss _ yy1 are positions of the upper left corner of the enhanced image in the number dimension in the original image, ss _ xx2 and ss _ yy2 are positions of the enhanced image in the lower right corner of the original image in the number dimension, and nw and nh are the width and the height of the enhanced image in the number dimension respectively;
step S23: judging whether the generated quantity dimension enhanced image contains a complete car logo area, if so, retaining the sub-image, otherwise, discarding;
step S24: a new emblem image is generated using the number dimension enhanced image positions calculated in step S22, and the calculated emblem position information is written into a file named by the original image file name plus the ss suffix.
4. The multi-dimensional emblem data enhancement method of claim 1, characterized in that: the step S3 specifically includes the following steps:
step S31: a function randn (0,1) is constructed to generate a number between 0 and 1 that should satisfy the normal distribution:
it is calculated that when μ is 0.5 and σ is 0.1938, the random number generated using the normal distribution can satisfy 0 to 1; the norm function in Matlab is used as a vector, so the randn (0,1) function is specifically constructed by using the norm function in Matlab (0.5, 0.1938);
step S32: calculating the width and height of the spatial dimension enhanced image:
wherein nw and nh are the width and height, respectively, of the spatial dimension enhanced image; ox and oy are the x coordinate and y coordinate of the center of the target area of the original image respectively; ow and oh are the width and height of the original image, respectively;
step S33: calculating the position of the central point of the car logo area in the generated space dimension enhanced image, wherein the calculation method comprises the following steps:
wherein nx and ny are respectively an x coordinate and a y coordinate of a target central point of the spatial dimension enhanced image; ow and oh are the width and height of the original image, respectively; nw and nh are the width and height, respectively, of the spatial dimension enhanced image; rw and rh are the width and height of the target region, respectively; randn (0,1) is a function that generates a number between 0 and 1 that satisfies a normal distribution;
step S34: and calculating the position of the generated space dimension enhanced image in the original image, wherein the calculation method comprises the following steps:
wherein mr _ xx1 and mr _ yy1 are positions of the upper left corner of the spatial dimension enhanced image in the original image; mr _ xx2 and mr _ yy2 are positions of the enhanced image in the lower right corner of the original image in the spatial dimension; ox and oy are the x coordinate and y coordinate of the center of the target area of the original image respectively; nx and ny are the x coordinate and the y coordinate of the target center point of the spatial dimension enhanced image, respectively; nw and nh are the width and height, respectively, of the spatial dimension enhanced image;
step S35: a new car logo image is generated using the spatial dimension enhanced image position calculated in step S34, and the calculated car logo position information is written in a file named by adding an mr suffix to the original image file name.
5. The multi-dimensional emblem data enhancement method of claim 1, characterized in that: the step S4 specifically includes the following steps:
step S41: integrating all the images generated in the steps S1, S2 and S3, and copying the original image so that the number ratio of the generated images to the original image is 3: 1;
step S42: performing brightness transformation processing on all the images generated in the steps S1, S2 and S3, wherein the brightness of each picture is randomly selected from 0.5 to 1.5 for transformation so as to strengthen the distribution of the data set in the brightness dimension;
step S43: performing contrast transformation processing on all the images generated in the steps S1, S2 and S3, wherein the contrast of each image is randomly selected from 0.5 to 1.5 for transformation, so as to strengthen the distribution of the data set in the contrast dimension;
step S44: all images are integrated to form a new data set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011114773.4A CN112200258B (en) | 2020-10-16 | 2020-10-16 | Multi-dimensional vehicle logo data enhancement method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011114773.4A CN112200258B (en) | 2020-10-16 | 2020-10-16 | Multi-dimensional vehicle logo data enhancement method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112200258A true CN112200258A (en) | 2021-01-08 |
| CN112200258B CN112200258B (en) | 2022-06-10 |
Family
ID=74010186
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011114773.4A Active CN112200258B (en) | 2020-10-16 | 2020-10-16 | Multi-dimensional vehicle logo data enhancement method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112200258B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN110096962A (en) * | 2019-04-04 | 2019-08-06 | 苏州千视通视觉科技股份有限公司 | Vehicle Detail based on region convolutional network identifies secondary structure method and device |
| US20200315587A1 (en) * | 2019-04-04 | 2020-10-08 | Koninklijke Philips N.V. | Data augmentation for training deep learning models with ultrasound images |
-
2020
- 2020-10-16 CN CN202011114773.4A patent/CN112200258B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN110096962A (en) * | 2019-04-04 | 2019-08-06 | 苏州千视通视觉科技股份有限公司 | Vehicle Detail based on region convolutional network identifies secondary structure method and device |
| US20200315587A1 (en) * | 2019-04-04 | 2020-10-08 | Koninklijke Philips N.V. | Data augmentation for training deep learning models with ultrasound images |
Non-Patent Citations (4)
| Title |
|---|
| MAHDI SALEH ET AL.: "Enhanced algorithm for autonomous parallel parking of a car-like mobile robot", 《2016 IEEE INTERNATIONAL MULTIDISCIPLINARY CONFERENCE ON ENGINEERING TECHNOLOGY (IMCET)》 * |
| 杨昌东等: "基于AT-PGGAN的增强数据车辆型号精细识别", 《中国图象图形学报》 * |
| 郭耀华: "基于深度学习的车辆驾驶状态识别算法研究", 《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》 * |
| 陈文兵等: "基于条件生成式对抗网络的数据增强方法", 《计算机应用》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112200258B (en) | 2022-06-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109003282B (en) | Image processing method and device and computer storage medium | |
| US7945114B2 (en) | Image transform method for obtaining expanded image data, image processing apparatus and image display device therefore | |
| TWI571803B (en) | Generation? method? of? color? qr? code | |
| KR100306250B1 (en) | Error concealer for video signal processor | |
| US7667778B2 (en) | Image processing apparatus and method, and recording medium and program used therewith | |
| US6686922B2 (en) | Adaptive thresholding using loose gray scale template matching | |
| US8265350B2 (en) | Method and apparatus for detecting and processing specific pattern from image | |
| US7848427B2 (en) | Apparatus and method for determining motion vector with effective pixel gradient | |
| US7813572B2 (en) | Compressing and restoring method of image data including a free microdot image element, a print dot image element, and a line picture image element | |
| CN110503146B (en) | Data enhancement method and device, computing equipment and computer storage medium | |
| CN113436112B (en) | Image enhancement method, device and equipment | |
| CN102257534A (en) | Method, apparatus and software for determining motion vectors | |
| JP2004166007A (en) | Device, method and program for image processing, and storage medium | |
| CN113592776A (en) | Image processing method and device, electronic device and storage medium | |
| US7327904B2 (en) | Pattern classification and filter design for increasing image resolution | |
| CN112200258B (en) | Multi-dimensional vehicle logo data enhancement method | |
| CN106127097B (en) | A kind of generation of the two dimensional code on prism surface and acquisition method | |
| Peng et al. | Efficient image resolution enhancement using edge-directed unsharp masking sharpening for real-time ASIC applications | |
| CN106446889B (en) | A kind of local recognition methods of logo and device | |
| CN106682670A (en) | Method and system for identifying station caption | |
| KR101287508B1 (en) | Apparatus and method for contrast enhancement using dominant brightness level and adaptive intensity transformation | |
| JP3026706B2 (en) | Image processing device | |
| CN109509237B (en) | Filter processing method and device and electronic equipment | |
| JP2004133592A (en) | Image processor, image processing method and image processing program for magnifying image | |
| CN112019700A (en) | Method for preventing secret-carrying image from being detected, intelligent terminal 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |