CN104732227B - A kind of Location Method of Vehicle License Plate based on definition and luminance evaluation - Google Patents
A kind of Location Method of Vehicle License Plate based on definition and luminance evaluation Download PDFInfo
- Publication number
- CN104732227B CN104732227B CN201510129964.0A CN201510129964A CN104732227B CN 104732227 B CN104732227 B CN 104732227B CN 201510129964 A CN201510129964 A CN 201510129964A CN 104732227 B CN104732227 B CN 104732227B
- Authority
- CN
- China
- Prior art keywords
- image
- definition
- license plate
- evaluation
- template
- 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.)
- Active
Links
Abstract
The invention discloses a kind of Location Method of Vehicle License Plate based on definition and luminance evaluation, comprise the steps：(1) intelligibility evaluation based on noise and acutance is carried out to input picture；(2) Grads Sharp processing is carried out if definition deficiency, Gaussian Blur processing is carried out if definition is too high；(3) gray-scale map is converted to by RGB figures；(4) luminance evaluation is carried out to gray-scale map；(5) illumination normalization is carried out if brightness is abnormal；(6) vertical edge in Scharr operator extraction images is passed through；(7) local auto-adaptive threshold process is carried out；(8) noise filtering is crossed by Morphological scale-space, and vertical edge is fused into connected region；(9) zone marker is carried out to connected region, is screened according to the feature of Chinese car plate and obtain license plate area.
Description
Technical Field
The invention belongs to the field of image processing, and particularly relates to a license plate rapid positioning method based on definition and brightness evaluation.
Background
In an intelligent traffic system, a license plate number recognition technology based on digital images becomes an indispensable important component and is widely applied to the fields of monitoring and alarming, violation punishment, access management, highway toll management and the like. The license plate positioning, namely positioning the position of the license plate in an image or a video, is a primary step of license plate number identification, the accuracy of the method directly influences the license plate number identification effect, and the method is the most time-consuming step. Therefore, the license plate positioning plays an important role in the intelligent transportation system.
Currently, license plate positioning methods are classified into three categories:
1. method based on edge information. The method is most widely applied and utilizes the characteristic that the vertical edge of the license plate is more to position. The method has good detection rate and real-time performance for simple scenes such as access management, toll stations and the like, but has low license plate positioning accuracy for complex scenes such as expressways and the like due to fuzzy license plates and uncontrollable illumination.
2. A method based on color information. The method utilizes the known license plate color information to assist the edge information of the license plate to position, can greatly improve the accuracy rate for the scene with single illumination condition, but can cause reaction in the scene with complex illumination such as night and the like, has higher requirement on the resolution of the image, and is only suitable for the license plate recognition system in the few realistic scenes.
3. A machine learning based method. The method is characterized in that license plate samples under different scenes are collected and used for training a global model containing different situations in the samples, and then the global model is popularized to unknown conditions in new images. The method has ideal effect when the definition or brightness of a new image is close to that of a training sample, but has low accuracy when the difference is large, and the real-time property is difficult to ensure.
In summary, the license plate location method based on the edge information is generally applied to a real-time automatic license plate recognition system due to the advantages of small computation amount, high processing speed, small required storage space, and the like, but the disadvantages are mainly embodied as the following two points: 1. the adjustment cannot be made according to the specific situation of input data; 2. the accuracy is low in a complex scene, and the real-time performance is poor.
Disclosure of Invention
The invention mainly aims to overcome the defects of the prior art and provide a license plate rapid positioning method based on definition and brightness evaluation, which effectively solves the problem that the license plate is difficult to be accurately positioned or cannot be positioned under different definitions and complex illumination environments, thereby improving the accuracy of license plate positioning.
In order to achieve the purpose, the invention adopts the following technical scheme:
a license plate rapid positioning method based on definition and brightness evaluation comprises the following steps:
(1) Performing a noise and sharpness based sharpness evaluation on the input image;
(2) Carrying out gradient sharpening on the image with insufficient definition, and carrying out Gaussian blur processing on the image with high definition;
(3) Converting the image processed in the step (2) into a gray scale image from an RGB image;
(4) Performing brightness evaluation on the gray level image, namely evaluating the gray level image under the conditions of over-brightness or under-brightness;
(5) If the brightness is abnormal, performing illumination normalization processing;
(6) Extracting a vertical edge in the image through a Scharr operator;
(7) Carrying out local adaptive threshold processing;
(8) Filtering noise through morphological processing, and fusing vertical edges into a connected region;
(9) And carrying out region marking on the connected region, and screening according to the characteristics of the Chinese license plate to obtain a license plate region.
Preferably, in step (1), the sharpness evaluation is composed of a noise evaluation and a sharpness evaluation, and after the noise evaluation and the sharpness evaluation of the image are obtained, the sharpness evaluation of the image is obtained through a certain weighting relationship, and the formula for calculating the sharpness evaluation is as follows:
Quality＝wNoise+(1-w)Sharpness
wherein w is a weight value, and the value range is between 0 and 1;
the noise evaluation is to obtain a binary image by performing OTSU threshold processing on an image, then calculate the ratio of the area in a communication area smaller than a given threshold, and finally obtain a noise evaluation value by linear transformation. The formula for calculating the noise estimate is as follows:
therein, comp i Is the ith Area, area i Is the area of the i-th region, T area A given threshold value for judging whether the area is too small, and a and b are parameters of linear transformation;
the sharpness evaluation adopts a point sharpness algorithm, each point of an image is subtracted by 8 neighboring points, the weighted sum of 8 difference values is firstly solved, the obtained values of all the points are added and divided by the total number of pixels, and the formula for calculating the sharpness evaluation value is as follows:
wherein a is the pixel of the a-th field, m and n are the length and width of the image, df is the change amplitude of the gray scale, and dx is the distance increment between the pixel points.
Preferably, the step (2) is specifically:
(2-1) given two thresholds T quality1 And T quality2 Wherein T is quality1 Is a threshold value for judging whether the image definition is insufficient, T quality2 Judging whether the image definition is over-high threshold value;
(2-2) when Quality < T quality1 Then, carrying out gradient sharpening on the image, namely solving gradient information Grad of the image through a Laplace operator, and then adding the gradient information Grad and the original image to obtain a gradient sharpening result, wherein the formula of the gradient sharpening is as follows:
f(x，y)＝f(x，y)+Grad(x，y)
wherein f (x, y) is the original image, and Grad (x, y) is gradient information;
(2-3) when Quality > T quality2 The image is subjected to Gaussian blur processing, namely a fuzzy template in normal distribution is calculated by using a Gaussian function, and convolution operation is carried out on the fuzzy template and the original image to obtain a Gaussian blur result, wherein the Gaussian function and the Gaussian blur formula are as follows:
f(x，y)＝f(x，y)*Gauss(a，b)
wherein, σ is the standard deviation of normal distribution, m and n are the length and width of the template, a and b are elements on the template, and f (x, y) is the original image.
Preferably, in step (3), the formula for converting the RGB map into the gray scale map is:
Gray(x，y)＝0.299R(x，y)+0.587G(x，y)+0.114B(x，y)；
or Gray (x, y) = (38R (x, y) +75G (x, y) +15B (x, y)) > 7;
orR, G, B is the value of the red, green, and blue color channels of the pixel (x, y), and Gray is the Gray value corresponding to the pixel (x, y).
Preferably, in the step (4), the method for evaluating the brightness of the gray scale is as follows:
calculating pixel values below a threshold T bright1 Or above the threshold T bright2 If it is greater than a given threshold value T bright3 The brightness is considered to be abnormal.
Preferably, in the step (5), the method for light normalization comprises:
(5-1) firstly, the brightness of the enhanced image is corrected by using Gamma, and the Gamma correction formula is as follows:
g(x，y)＝c[f(x，y)]γ
wherein f (x, y) is an input image, g (x, y) is an output image, c is an arbitrary value for adjusting the contrast of the image, and γ is between 0 and 1 for adjusting the brightness of the image;
(5-2) then, gaussian difference filtering is carried out to remove high-frequency components, namely uneven parts, in the illumination, and the Gaussian difference filtering formula is as follows:
g(x，y)＝f(x，y)*Gauss(σ 2 )-f(x，y)*Gauss(σ 1 )
where f (x, y) is the input image, g (x, y) is the output image, gauss (σ) n ) Is that the variance is σ n (n =1,2);
(5-3) finally, contrast equalization is adopted to enhance the contrast of the image illumination, so that unevenness in the illumination is removed, meanwhile, useful details of the image are retained to the maximum degree, and the image contrast is enhanced, wherein the contrast equalization formula is as follows:
where f (x, y) is the input image, g n (x, y) (n =1,2,3) is the output image, α is the compressibility index for reducing the effect of large pixel values, and τ is the large pixel value in the truncated image after the first normalization.
Preferably, in the step (6), the specific method for extracting the vertical edge in the image through the Scharr operator is as follows:
the Scharr template is expressed as follows:
followed by Grad x The image is convolved to obtain the vertical edge, and the vertical edge extraction formula is as follows:
f(x，y)＝f(x，y)*Grad x
where f (x, y) is the original.
Preferably, in step (7), the specific method for performing the local adaptive threshold processing is as follows:
(7-1) dividing the image into m × n small areas of equal size;
(7-2) carrying out OTSU processing in each cell, and recording a two-dimensional matrix composed of all threshold values as Mat _ T 0 ；
(7-3) for all Mat _ T 0 Output Mat _ T by performing 3 × 3 Gaussian filtering 1 Then according to Mat _ T 1 Image binarization for each region is performed.
Preferably, the step (8) is specifically:
(8-1) performing noise filtering on the image through a morphological opening operation, specifically, constructing an m 1 ×n 1 Then, the image is corroded firstly and then expanded;
(8-2) fusing vertical edges in the image into a connected region through a morphological closing operation, and specifically constructing an m 2 ×n 2 Then, the image is subjected to the operation of expansion and corrosion;
the etching is to etch the edge of the object, and the specific process is to perform the following processing on each pixel (x, y) in the image by using the template: the pixel (x, y) is placed in the center of the template and the value of the pixel (x, y) is modified to the smallest value of all pixels over all other pixels covered by the template according to the template size. The expansion is opposite to the erosion, and the outline of the object is expanded, and the specific process is that the template performs the following processing on each pixel (x, y) in the image: the pixel (x, y) is placed in the center of the template and, depending on the size of the template, the value of all pixels in the template is modified to the largest of all pixels over all other pixels covered by the template.
Preferably, the step (9) is specifically:
(9-1) carrying out region marking on the communicated region, and calculating the minimum circumscribed rectangle of the communicated region;
(9-2) according to the ASPECT ratio ASPECT and the AREA size AREA of the Chinese license plate, giving a deviation rate e of the ASPECT ratio 1 Deviation ratio e of sum area 2 The aspect ratio aspect and the area of the minimum bounding rectangle of each connected region are subjected to the following tests:
|aspect-ASPECT|≤e 1 ASPECT
|area-AREA|≤e 2 AREA
and (4) carrying out the step (9-3) on the connected region meeting the conditions, or else, carrying out the non-license plate region;
(9-3) in general, the inclination angle of the license plate in the image is not too large, so that a threshold value T of the inclination angle is given angle The inclination angle of each connected region is checked by:
|angle-0|≤T angle
and if the connected region meets the conditions, the connected region is regarded as a license plate region, otherwise, the connected region is a non-license plate region.
Compared with the prior art, the invention has the following advantages and beneficial effects:
1. according to the invention, the Scharr operator is used for replacing the Sobel operator to carry out edge detection, so that the vertical edge of the license plate can be detected more accurately, and the same high efficiency is maintained.
2. The invention innovatively introduces the definition evaluation module to adaptively adjust the input images with different definitions, is more beneficial to subsequent processing and greatly improves the accuracy of license plate positioning under different definitions.
3. The invention innovatively introduces the brightness evaluation module to perform illumination normalization processing on the input image with abnormal brightness, so that the subsequent processing is facilitated, and the accuracy of license plate positioning in a complex illumination environment is greatly improved.
Drawings
Fig. 1 is a general flowchart of a license plate quick positioning method based on definition and brightness evaluation according to the present invention.
Detailed Description
The present invention will be described in further detail with reference to examples and drawings, but the present invention is not limited thereto.
Examples
As shown in fig. 1, the license plate fast positioning method based on the definition and brightness evaluation of the embodiment specifically includes the following steps:
1. noise and sharpness based sharpness evaluation of the input images:
(1) the noise evaluation is to obtain a binary image by performing OTSU threshold processing on the image, then calculate the ratio of the area in the communication area smaller than a given threshold, and finally obtain the noise evaluation value by linear transformation. The formula for calculating the noise estimate is as follows:
therein, comp i Is the ith Area, area i Is the area of the i-th region, T area Is a given threshold to determine if the area is too small, and a and b are parameters of the linear transformation.
(2) Sharpness evaluation adopts a point sharpness algorithm, each point of an image is subtracted from 8 neighboring points, and a weighted sum of 8 difference values is firstly obtained (the weight depends on the distance, if the distance is long, the weight is small, and if the difference values in the directions of 45 degrees and 135 degrees need to be multiplied) The value obtained at all points is added up and divided by the total number of pixels. The formula for calculating the sharpness evaluation value is as follows:
wherein a is the pixel of the a-th field, m and n are the length and width of the image, df is the change amplitude of the gray scale, and dx is the distance increment between the pixel points.
(3) The definition evaluation is composed of noise evaluation and sharpness evaluation, and after the noise evaluation value and the sharpness evaluation value of the image are obtained, the definition evaluation value of the image is obtained through a certain weighting relation. The formula for calculating the sharpness evaluation value is as follows:
Quality＝wNoise+(1-w)Sharpness
wherein w is a weight value, and the value range is between 0 and 1.
2. Carrying out gradient sharpening on the image with insufficient definition, and carrying out Gaussian blur processing on the image with high definition:
(1) given two thresholds T quality1 And T quality2 Wherein T is quality1 Is a threshold value for judging whether the image definition is insufficient, T quality2 Is a threshold value for judging whether the image definition is too high.
(2) When Quality < T quality1 Then, carrying out gradient sharpening on the image, namely solving gradient information Grad of the image through a Laplace operator, and then adding the gradient information Grad and the original image to obtain a gradient sharpening result, wherein the formula of the gradient sharpening is as follows:
f(x，y)＝f(x，y)+Grad(x，y)
where f (x, y) is the original image and Grad (x, y) is the gradient information.
(3) When Quality > T quality2 The image is subjected to Gaussian blur processing, namely a fuzzy template in normal distribution is calculated by using a Gaussian function, and convolution operation is carried out on the fuzzy template and the original image to obtain a Gaussian blur result, wherein the Gaussian function and the Gaussian blur formula are as follows:
f(x，y)＝f(x，y)*Gauss(a，b)
wherein, σ is the standard deviation of normal distribution, m and n are the length and width of the template, a and b are elements on the template, and f (x, y) is the original image.
3. Converting the RGB image into a gray scale image, wherein the conversion formula is as follows:
Gray(x，y)＝0.299R(x，y)+0.587G(x，y)+0.114B(x，y)
Gray(x，y)＝(38R(x，y)+75G(x，y)+15B(x，y))＞＞7
r, G, B is the value of the red, green, and blue color channels of the pixel (x, y), and Gray is the Gray value corresponding to the pixel (x, y).
4. And (3) evaluating the brightness of the gray level image:
evaluating the gray image under the condition of over-brightness or under-brightness, specifically calculating the pixel value below the threshold value T bright1 Or above the threshold T bright2 If it is greater than a given threshold value T bright3 Then it is recognizedIf so, it is a luminance abnormality.
5. And if the brightness is abnormal, performing illumination normalization:
(1) firstly, gamma correction is used for enhancing the image brightness, and the Gamma correction formula is as follows:
g(x，y)＝c[f(x，y)] γ
wherein f (x, y) is an input image, g (x, y) is an output image, c is an arbitrary value for adjusting the contrast of the image, and γ is between 0 and 1 for adjusting the brightness of the image;
(2) then, gaussian difference filtering (DOG, difference of Gaussian) is performed to remove high-frequency components, i.e. non-uniform parts, in the illumination, and the DOG filtering formula is as follows:
g(x，y)＝f(x，y)*Gauss(σ 2 )-f(x，y)*Gauss(σ 1 )
where f (x, y) is the input image, g (x, y) is the output image, gauss (σ) n ) Is that the variance is σ n (n =1,2);
(3) and finally, contrast equalization is adopted to enhance the contrast of the image illumination, so that unevenness in the illumination is removed, simultaneously, useful details of the image are retained to the maximum degree, and the contrast of the image is enhanced, wherein the contrast equalization formula is as follows:
where f (x, y) is the input image, g n (x, y) (n =1,2,3) is the output image, α is the compressibility index for reducing the effect of large pixel values, τ is used to truncate large pixel values in the image after the first step of normalization;
6. extracting vertical edges in the image through a Scharr operator:
in order to extract image edge information, a Sobel operator is commonly used, but since the Sobel operator is often not very accurate in calculation on a 3 × 3 convolution template, an improved Sobel operator is called a Scharr operator, the operation efficiency of the improved Sobel operator is as fast as that of the Sobel operator, but the result is more accurate, and the Scharr template is expressed as follows:
since the horizontal edge is generally not beneficial but harmful to the license plate positioning, grad is only adopted here x The image is convoluted to obtain the vertical edge of the image, and the vertical edge extraction formula is as follows:
f(x，y)＝f(x，y)*Grad x
where f (x, y) is the original.
7. And (3) local adaptive threshold processing is carried out:
(1) dividing the image into m × n small regions of equal size;
(2) performing OTSU processing in each cell, and recording a two-dimensional matrix composed of all threshold values as Mat _ T 0 ；
(3) For all Mat _ T 0 Output Mat _ T by performing 3 × 3 Gaussian filtering 1 Then according to Mat _ T 1 Carrying out image binarization on each region;
8. filtering noise by morphological processing, and fusing vertical edges into connected regions:
(1) performing noise filtering on the image by morphological opening operation, specifically constructing an m 1 ×n 1 Then, the image is corroded firstly and then expanded;
(2) fusing vertical edges in the image into a connected region through morphological closing operation, specifically constructing an m 2 ×n 2 Then, the image is subjected to the operation of expansion and corrosion;
the etching is to etch the edge of the object, and the specific process is to perform the following processing on each pixel (x, y) in the image by using the template: the pixel (x, y) is placed in the center of the template and the value of the pixel (x, y) is modified to the smallest value of all pixels over all other pixels covered by the template according to the template size. The expansion is opposite to the erosion, and the outline of the object is expanded, and the specific process is that the template performs the following processing on each pixel (x, y) in the image: the pixel (x, y) is placed in the center of the template and, depending on the size of the template, all pixels within the template are modified to the largest of all pixels over all other pixels covered by the template.
9. And (3) carrying out region marking on the connected region, screening according to the characteristics of the Chinese license plate and obtaining a license plate region:
(1) carrying out region marking on the connected region, and calculating the minimum external rectangle of the connected region;
(2) according to the ASPECT ratio ASPECT and the AREA size AREA of the Chinese license plate, a deviation rate e of the ASPECT ratio is given 1 Deviation ratio e of sum area 2 The aspect ratio aspect and the area of the minimum bounding rectangle of each connected region are subjected to the following tests:
|aspect-ASPECT|≤e 1 ASPECT
|area-AREA|≤e 2 AREA
and (4) carrying out the step (3) on the connected region meeting the conditions, otherwise, the connected region is a non-license plate region.
(3) Generally, the inclination angle of the license plate in the image is not too large, so a threshold value T of the inclination angle is given angle The inclination angle of each connected region is checked by:
|angle-0|≤T angle
and if the connected region meets the conditions, the connected region is regarded as a license plate region, otherwise, the connected region is a non-license plate region.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.
Claims (9)
1. A license plate rapid positioning method based on definition and brightness evaluation is characterized by comprising the following steps:
(1) Performing a sharpness evaluation based on noise and sharpness on the input image; the definition evaluation is composed of a noise evaluation and a sharpness evaluation, after the noise evaluation value and the sharpness evaluation value of the image are obtained, the definition evaluation value of the image is obtained through a weighting relation, and a formula for calculating the definition evaluation value is as follows:
Quality＝wNoise+(1-w)Sharpness
wherein w is a weight value, and the value range is between 0 and 1;
the noise evaluation is to obtain a binary image by performing OTSU threshold processing on an image, then calculate the occupation ratio of a communication area smaller than a given threshold, and finally obtain a noise evaluation value by linear transformation, wherein the formula for calculating the noise evaluation value is as follows:
therein, comp i Is the ith Area, area i Is the area of the i-th region, T area Is a given threshold value, L, for judging whether the area is too small 1 And L 2 Is a parameter of the linear transformation;
the sharpness evaluation adopts a point sharpness algorithm, each point of an image is subtracted by 8 neighboring points, the weighted sum of 8 difference values is firstly solved, then the obtained values of all the points are added and divided by the total number of pixels, and the formula for calculating the sharpness evaluation value is as follows:
wherein k is the kth neighborhood pixel, m and n are the length and width of the image, df is the change amplitude of the gray scale, and dx is the distance increment between the pixel points;
(2) Carrying out gradient sharpening on the image with insufficient definition, and carrying out Gaussian blur processing on the image with high definition;
(3) Converting the image processed in the step (2) into a gray scale image from an RGB image;
(4) Performing brightness evaluation on the gray level image, namely evaluating the gray level image under the conditions of over-brightness or under-brightness;
(5) If the brightness is abnormal, performing illumination normalization processing;
(6) Extracting a vertical edge in the image through a Scharr operator;
(7) Carrying out local adaptive threshold processing;
(8) Filtering noise through morphological processing, and fusing vertical edges into a connected region;
(9) And carrying out region marking on the connected region, and screening according to the characteristics of the Chinese license plate to obtain a license plate region.
2. The license plate rapid positioning method based on definition and brightness evaluation according to claim 1, wherein the step (2) is specifically:
(2-1) given two thresholds T quality1 And T quality2 Wherein T is quality1 Is a threshold value for judging whether the image definition is insufficient, T quality2 Judging whether the image definition is over-high threshold value;
(2-2) when Quality<T quality1 Then, carrying out gradient sharpening on the image, namely solving gradient information Grad of the image through a Laplace operator, and then adding the gradient information Grad and the original image to obtain a gradient sharpening result, wherein the formula of the gradient sharpening is as follows:
f1(x,y)＝f(x,y)+Grad(x,y)
wherein f (x, y) is the original image, f1 (x, y) is the image obtained after the original image is sharpened, and Grad (x, y) is gradient information;
(2-3) when Quality>T quality2 Then, the image is processed with Gaussian blur, namely, a blur template in normal distribution is calculated by using a Gaussian function, and convolution operation is carried out on the blur template and the original image to obtain a Gaussian blur result, namely, a Gaussian functionThe equations for number and gaussian blur are as follows:
f2(x,y)＝f(x,y)*Gauss(a,b)
wherein σ is a standard deviation of normal distribution, m and n are template length and width, a and b are elements on the template, f (x, y) is original image, and f2 (x, y) is image obtained by Gaussian blurring of the original image.
3. The license plate rapid positioning method based on the definition and brightness evaluation as claimed in claim 1, wherein in the step (3), the formula of converting the RGB image into the gray scale image is:
Gray(x,y)＝0.299R(x,y)+0.587G(x,y)+0.114B(x,y)；
or Gray (x, y) = (38R (x, y) +75G (x, y) +15B (x, y)) > 7;
orR, G, B is the value of the red, green, and blue color channels of the pixel (x, y), and Gray is the Gray value corresponding to the pixel (x, y).
4. The license plate rapid positioning method based on definition and brightness evaluation as claimed in claim 1, wherein in the step (4), the brightness evaluation method for the gray scale map comprises:
calculating pixel values below a threshold T bright1 Or above the threshold T bright2 If it is greater than a given threshold value T bright3 It is considered to be a luminance abnormality.
5. The license plate rapid positioning method based on the definition and brightness evaluation as claimed in claim 1, wherein in the step (5), the illumination normalization processing method comprises:
(5-1) firstly, the brightness of the enhanced image is corrected by using Gamma, and the Gamma correction formula is as follows:
g(x,y)＝c[f(x,y)] γ
wherein f (x, y) is an input image, g (x, y) is an output image, c is an arbitrary value for adjusting the contrast of the image, and γ is between 0 and 1 for adjusting the brightness of the image;
(5-2) Gaussian difference filtering is then carried out to remove high-frequency components, namely uneven parts, in the illumination, and the Gaussian difference filtering formula is as follows:
g(x,y)＝f(x,y)*Gauss(σ 2 )-f(x,y)*Gauss(σ 1 )
where f (x, y) is the input image, g (x, y) is the output image, gauss (σ) n ) Is that the variance is σ n (n =1,2);
(5-3) finally, contrast equalization is adopted to enhance the contrast of the image illumination, so that unevenness in the illumination is removed, meanwhile, useful details of the image are retained to the maximum degree, and the image contrast is enhanced, wherein the contrast equalization formula is as follows:
where f (x, y) is the input image, g n (x, y) (n =1,2,3) is the output image, α is the compressibility index used to reduce the effect of large pixel values, and τ is used to truncate large pixel values in the image after the first step of normalization.
6. The license plate rapid positioning method based on the definition and brightness evaluation as claimed in claim 1, wherein in the step (6), the specific method for extracting the vertical edge in the image by the Scharr operator is as follows:
the Scharr template is expressed as follows:
followed by Grad x The image is convoluted to obtain the vertical edge of the image, and the vertical edge extraction formula is as follows:
f3(x,y)＝f(x,y)*Grad x
wherein f (x, y) is the original image, and f3 (x, y) is the image extracted from the vertical edge of the original image.
7. The license plate rapid positioning method based on the definition and brightness evaluation as claimed in claim 1, wherein in the step (7), the specific method for performing the local adaptive threshold processing is as follows:
(7-1) dividing the image into m × n small areas of equal size;
(7-2) carrying out OTSU processing in each cell, and recording a two-dimensional matrix composed of all threshold values as Mat _ T 0 ；
(7-3) for all Mat _ Ts 0 Output Mat _ T by performing 3 × 3 Gaussian filtering 1 Then according to Mat _ T 1 Image binarization for each region is performed.
8. The license plate rapid positioning method based on the definition and brightness evaluation as claimed in claim 1, wherein the step (8) is specifically as follows:
(8-1) performing noise filtering on the image through a morphological opening operation, specifically, constructing an m 1 ×n 1 Then, the image is corroded firstly and then expanded;
(8-2) fusing vertical edges in the image into a connected region through a morphological closing operation, and specifically constructing an m 2 ×n 2 Then, the image is subjected to the operation of expansion and corrosion;
the etching is to etch the edge of the object, and the specific process is to perform the following processing on each pixel (x, y) in the image by using the template: placing the pixel (x, y) in the center of the template, traversing all other pixels covered by the template according to the size of the template, modifying the value of the pixel (x, y) to be the minimum value of all pixels, expanding the outline of the object as opposed to eroding, and specifically processing each pixel (x, y) in the image by the template as follows: the pixel (x, y) is placed in the center of the template and, depending on the size of the template, all pixels within the template are modified to the largest of all pixels over all other pixels covered by the template.
9. The license plate rapid positioning method based on definition and brightness evaluation according to claim 1, wherein the step (9) is specifically:
(9-1) carrying out region marking on the connected region, and calculating the minimum circumscribed rectangle of the connected region;
(9-2) according to the ASPECT ratio ASPECT and the AREA size AREA of the Chinese license plate, giving a deviation rate e of the ASPECT ratio 1 Deviation ratio of sum area e 2 The aspect ratio aspect and the area of the minimum bounding rectangle of each connected region are subjected to the following tests:
|aspect-ASPECT|≤e 1 ASPECT ①
|area-AREA|≤e 2 AREA ②
carrying out the step (9-3) on a communication area which simultaneously satisfies the formulas (1) and (2), otherwise, a non-license plate area;
(9-3) given a threshold value T of the inclination angle angle The inclination angle of each connected region is checked by:
|angle-0|≤T angle ③
and (4) if the connected region satisfying the formula (3) is a license plate region, otherwise, the connected region is a non-license plate region.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510129964.0A CN104732227B (en) | 2015-03-23 | 2015-03-23 | A kind of Location Method of Vehicle License Plate based on definition and luminance evaluation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510129964.0A CN104732227B (en) | 2015-03-23 | 2015-03-23 | A kind of Location Method of Vehicle License Plate based on definition and luminance evaluation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104732227A CN104732227A (en) | 2015-06-24 |
| CN104732227B true CN104732227B (en) | 2017-12-26 |
Family
ID=53456102
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510129964.0A Active CN104732227B (en) | 2015-03-23 | 2015-03-23 | A kind of Location Method of Vehicle License Plate based on definition and luminance evaluation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104732227B (en) |
Families Citing this family (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10706330B2 (en) * | 2015-10-01 | 2020-07-07 | Intellivision Technologies Corp | Methods and systems for accurately recognizing vehicle license plates |
| US11587327B2 (en) | 2015-10-01 | 2023-02-21 | Intellivision Technologies Corp | Methods and systems for accurately recognizing vehicle license plates |
| CN105898174A (en) * | 2015-12-04 | 2016-08-24 | 乐视网信息技术（北京）股份有限公司 | Video resolution improving method and device |
| JP2018091807A (en) * | 2016-12-07 | 2018-06-14 | オルボテック リミテッド | Defective flaw determination method and device |
| CN107240092B (en) * | 2017-05-05 | 2020-02-14 | 浙江大华技术股份有限公司 | Image ambiguity detection method and device |
| CN109584198B (en) * | 2017-09-26 | 2022-12-23 | 浙江宇视科技有限公司 | Method and device for evaluating quality of face image and computer readable storage medium |
| CN108460766B (en) * | 2018-04-12 | 2022-02-25 | 四川和生视界医药技术开发有限公司 | Retina image definition evaluation method and evaluation device |
| CN108550158B (en) * | 2018-04-16 | 2021-12-17 | Tcl华星光电技术有限公司 | Image edge processing method, electronic device and computer readable storage medium |
| CN108710852B (en) * | 2018-05-21 | 2021-08-03 | 山东大学 | Particle size distribution image recognition method and system for limiting shooting depth |
| CN109915924B (en) * | 2018-05-31 | 2020-11-24 | 徐州云创物业服务有限公司 | Safety protection type electric heater |
| CN109509229B (en) * | 2018-11-12 | 2020-12-15 | 凌云光技术集团有限责任公司 | Template reconstruction device and method based on two-dimensional linear transformation |
| CN109394268B (en) * | 2018-12-07 | 2021-05-11 | 刘志红 | Polyp harm degree mapping platform |
| CN110111261B (en) * | 2019-03-28 | 2021-05-28 | 瑞芯微电子股份有限公司 | Adaptive balance processing method for image, electronic device and computer readable storage medium |
| CN110400261A (en) * | 2019-04-04 | 2019-11-01 | 桑尼环保（江苏）有限公司 | Adaptive environment clears up platform |
| CN111027363B8 (en) * | 2019-04-28 | 2021-03-19 | 丽水新贝蕾科技有限公司 | Auxiliary signal parameter analysis system |
| CN110246227B (en) * | 2019-05-21 | 2023-12-29 | 佛山科学技术学院 | Virtual-real fusion simulation experiment image data collection method and system |
| CN110544229B (en) * | 2019-07-11 | 2022-11-18 | 华南理工大学 | Image focus evaluation and focusing method under non-uniform speed state of camera |
| CN111353994B (en) * | 2020-03-30 | 2023-06-30 | 南京工程学院 | Image non-reference brightness quality detection method for target detection |
| CN111611863B (en) * | 2020-04-22 | 2022-05-20 | 浙江大华技术股份有限公司 | License plate image quality evaluation method and device and computer equipment |
| CN112258503B (en) * | 2020-11-13 | 2023-11-14 | 中国科学院深圳先进技术研究院 | Ultrasonic image imaging quality evaluation method, device and computer readable storage medium |
| CN112863010B (en) * | 2020-12-29 | 2022-08-05 | 宁波友好智能安防科技有限公司 | Video image processing system of anti-theft lock |
| CN114764775A (en) * | 2021-01-12 | 2022-07-19 | 深圳市普渡科技有限公司 | Infrared image quality evaluation method, device and storage medium |
| CN113486895B (en) * | 2021-07-16 | 2022-04-19 | 浙江华是科技股份有限公司 | Ship board identification method and device and computer readable storage medium |
| CN117173416B (en) * | 2023-11-01 | 2024-01-05 | 山西阳光三极科技股份有限公司 | Railway freight train number image definition processing method based on image processing |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103093229A (en) * | 2013-01-21 | 2013-05-08 | 信帧电子技术（北京）有限公司 | Positioning method and device of vehicle logo |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8331621B1 (en) * | 2001-10-17 | 2012-12-11 | United Toll Systems, Inc. | Vehicle image capture system |
-
2015
- 2015-03-23 CN CN201510129964.0A patent/CN104732227B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103093229A (en) * | 2013-01-21 | 2013-05-08 | 信帧电子技术（北京）有限公司 | Positioning method and device of vehicle logo |
Non-Patent Citations (1)
| Title |
|---|
| 复杂光照条件下的通用车牌定位系统的研究与实现;刘攀;《中国优秀硕士学位论文全文数据库信息科技辑》;20131215(第S2期);第6、11-12、14-15、17-19、31-32、54、59、64-66页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104732227A (en) | 2015-06-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104732227B (en) | A kind of Location Method of Vehicle License Plate based on definition and luminance evaluation | |
| CN106651872B (en) | Pavement crack identification method and system based on Prewitt operator | |
| CN109886896B (en) | Blue license plate segmentation and correction method | |
| CN103543394B (en) | A kind of high voltage electric equipment electric discharge ultraviolet imagery quantization parameter extracting method | |
| CN102063706A (en) | Rapid defogging method | |
| CN107705254B (en) | City environment assessment method based on street view | |
| CN102306307B (en) | Positioning method of fixed point noise in color microscopic image sequence | |
| CN108090459B (en) | Traffic sign detection and identification method suitable for vehicle-mounted vision system | |
| CN103413311A (en) | Edge-based fuzzy detection method | |
| CN107742291A (en) | A kind of the defects of photovoltaic glass detection method and device | |
| CN103698676A (en) | Method and system for evaluating corona discharge of electric power transmission and transformation device | |
| CN116630813B (en) | Highway road surface construction quality intelligent detection system | |
| CN115131375B (en) | Automatic ore cutting method | |
| CN112862832A (en) | Dirt detection method based on concentric circle segmentation positioning | |
| CN117094975A (en) | Method and device for detecting surface defects of steel and electronic equipment | |
| CN116524196A (en) | Intelligent power transmission line detection system based on image recognition technology | |
| CN111652033A (en) | Lane line detection method based on OpenCV | |
| CN111241911B (en) | Self-adaptive lane line detection method | |
| CN109800693B (en) | Night vehicle detection method based on color channel mixing characteristics | |
| CN108898080B (en) | Ridge line neighborhood evaluation model-based crack connection method | |
| CN107154045B (en) | Accurate positioning method for vehicles on flow guide area illegal rolling line based on computer vision | |
| CN114120061A (en) | Small target defect detection method and system for power inspection scene | |
| CN104899898A (en) | Multidimensional information probability model based road surface detection method | |
| CN108230258B (en) | License plate region enhancement method based on horizontal neighborhood standard deviation calculation | |
| CN114612333A (en) | Microscopic image-based microencapsulated algae cell statistical method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |