CN104732227B - A kind of Location Method of Vehicle License Plate based on definition and luminance evaluation - Google Patents

Abstract

The invention discloses a method for quickly locating a license plate based on sharpness and brightness evaluation, which comprises the following steps: (1) carrying out sharpness evaluation based on noise and sharpness on an input image; (2) performing a gradient if the sharpness is insufficient Sharpening processing, if the clarity is too high, perform Gaussian blur processing; (3) convert from RGB image to grayscale image; (4) evaluate the brightness of the grayscale image; (5) perform light normalization processing if the brightness is abnormal ; (6) Extract vertical edges in the image through the Scharr operator; (7) Perform local adaptive threshold processing; (8) Filter noise through morphological processing, and make vertical edges merge into connected regions; (9) Connect the connected regions Area mark, filter according to the characteristics of the Chinese license plate and get the license plate area.

Description

A Fast Locating Method of License Plate Based on Clarity and Brightness Evaluation

technical field

The invention belongs to the field of image processing, in particular to a method for quickly locating a license plate based on definition and brightness evaluation.

Background technique

In the intelligent transportation system, the license plate number recognition technology based on digital images has become an indispensable and important part and is widely used in the fields of monitoring and alarming, punishment for violations, access management and highway toll management. License plate location, that is, locating the position of the license plate in the image or video, is the first step in license plate number recognition. Its accuracy directly affects the effect of license plate number recognition, and it is also the most time-consuming step. Therefore, license plate location plays an important role in intelligent transportation systems.

Currently, license plate positioning methods are divided into three categories:

1. Method based on edge information. This method is the most widely used, and uses the features with more vertical edges of the license plate for positioning. For simple scenes such as access management and toll booths, it has good detection rate and real-time performance, but for complex scenes such as expressways, the accuracy of license plate location is low due to the blurred license plate and uncontrollable lighting.

2. Method based on color information. This method uses the known color information of the license plate to assist the edge information of the license plate to locate. It can greatly improve the accuracy rate for scenes with single lighting conditions, but it will cause adverse effects in complex lighting scenarios such as night, and the image resolution requirements are also high. Higher, only suitable for license plate recognition systems in very few realistic scenarios.

3. Method based on machine learning. This method collects license plate samples in different scenes and uses them to train a global model that includes different situations in the samples, and then generalizes to unknown situations in new images. When the clarity or brightness of the new image is close to the training sample, the effect is ideal, but when the difference is large, the accuracy is low and it is difficult to guarantee real-time performance.

To sum up, the license plate location method based on edge information is widely used in the real-time automatic license plate recognition system due to the advantages of small amount of calculation, fast processing speed, and small required storage space, but its shortcomings are mainly reflected in the following two points: 1. 1. It cannot be adjusted according to the specific situation of the input data; 2. In complex scenes, the accuracy rate is low and the real-time performance is poor.

Contents of the invention

The main purpose of the present invention is to overcome the shortcomings and deficiencies of the prior art, and provide a method for quickly locating license plates based on evaluation of clarity and brightness, which effectively solves the problem that license plates are difficult to locate accurately under different clarity and complex lighting environments. Or can not locate the problem, thereby improving the accuracy of the license plate location.

In order to achieve the above object, the present invention adopts the following technical solutions:

A method for quickly locating license plates based on clarity and brightness evaluation, comprising the following steps:

(1) Carry out sharpness evaluation based on noise and sharpness to the input image;

(2) Gradient sharpening processing is performed on images with insufficient clarity, and Gaussian blur processing is performed on images with excessive clarity;

(3) the image processed in step (2) is converted into a grayscale image by the RGB image;

(4) Evaluate the brightness of the grayscale image, that is, evaluate the grayscale image in two situations of excessive brightness or insufficient brightness;

(5) If the brightness is abnormal, normalize the illumination;

(6) extract the vertical edge in the image by the Scharr operator;

(7) Perform local adaptive threshold processing;

(8) filter noise through morphological processing, and merge vertical edges into connected regions;

(9) Carry out area marking on the connected area, screen according to the characteristics of the Chinese license plate and obtain the license plate area.

Preferably, in step (1), the sharpness evaluation is composed of noise evaluation and sharpness evaluation. After obtaining the noise evaluation value and sharpness evaluation value of the image, the sharpness evaluation value of the image is obtained through a certain weighted relationship, and the sharpness evaluation is calculated. The formula for degree evaluation value is as follows:

Quality=wNoise+(1-w)Sharpness

Among them, w is the 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 the image first, then calculate the proportion of connected regions whose area is smaller than a given threshold, and finally obtain the noise evaluation value through linear transformation. The formula for calculating the noise evaluation value is as follows:

Among them, Comp _i is the i-th area, Area _i is the area of the i-th area, T _area is a given threshold for judging whether the area is too small, a and b are the parameters of the linear transformation;

The sharpness evaluation adopts the point sharpness algorithm, and subtracts 8 neighborhood points from each point of the image, and first calculates the weighted sum of the 8 differences, and then adds and divides the obtained values of all points by the total pixel value. The formula for calculating the sharpness evaluation value is as follows:

Among them, a is the pixel in the ath area, m and n are the length and width of the image, df is the change amplitude of the gray level, and dx is the distance increment between pixels.

Preferably, step (2) is specifically:

(2-1) Two thresholds T _quality1 and T _quality2 are given, wherein T _quality1 is the threshold for judging whether the image definition is insufficient, and T _quality2 is the threshold for judging whether the image definition is too high;

(2-2) When Quality<T _quality1 , the gradient sharpening process is performed on the image, that is, the image gradient information Grad is obtained through the Laplace operator, and then added to the original image to obtain the gradient sharpening result. The gradient sharpening formula is as follows:

f(x,y)=f(x,y)+Grad(x,y)

Among them, f(x, y) is the original image, Grad(x, y) is the gradient information;

(2-3) When Quality>T _quality2 , Gaussian blur processing is performed on the image, that is, the Gaussian function is used to calculate the blur template with a normal distribution, and the Gaussian blur result is obtained by convolution with the original image, Gaussian function and Gaussian blur The formula is as follows:

f(x,y)=f(x,y)*Gauss(a,b)

Among them, σ is the standard deviation of the normal distribution, m and n are the length and width of the template, a and b are the elements on the template, and f(x, y) is the original image.

Preferably, in step (3), the formula of converting the RGB image to the grayscale 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;

or Among them, R, G, and B are the values 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 step (4), the method for carrying out brightness evaluation to grayscale is:

Calculate the sum of the proportions of pixels whose pixel value is lower than the threshold T _bright1 or higher than the threshold T _bright2 to all pixels, and if it is greater than the given threshold T _bright3 , it is considered to be an abnormal brightness.

Preferably, in step (5), the method for normalizing the illumination is:

(5-1) First use Gamma correction to enhance the brightness of the image. The Gamma correction formula is as follows:

g(x,y)=c[f(x,y)]γ

In the formula, f(x, y) is the input image, g(x, y) is the output image, c is any value used to adjust the image contrast, and γ is used to adjust the image brightness between 0 and 1;

(5-2) Then perform Gaussian difference filtering to remove the high-frequency component of the illumination, that is, the uneven part. The Gaussian difference filtering formula is as follows:

g(x,y)=f(x,y)*Gauss(σ ₂ )-f(x,y)*Gauss(σ ₁ )

In the formula, f(x, y) is an input image, g(x, y) is an output image, and Gauss(σ _n ) is a Gaussian function whose variance is σ _n (n=1, 2);

(5-3) Finally, the contrast equalization is used to enhance the contrast of the image illumination, so as to remove the unevenness of the illumination while retaining the useful details of the image to the greatest extent and enhance the image contrast. 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 influence of large pixel values, τ is used for the first Cut off large pixel values in the image after one-step normalization.

Preferably, in step (6), the specific method of extracting the vertical edge in the image by the Scharr operator is:

The Scharr template expression is as follows:

Then use Grad _x to convolve the image to obtain its vertical edge, and the vertical edge extraction formula is as follows:

f(x,y)=f(x,y)*Grad _x

Among them, f(x, y) is the original image.

Preferably, in step (7), the specific method for performing local adaptive threshold processing is:

(7-1) Divide the image into m×n small areas of equal size;

(7-2) Carry out OTSU processing in each small area, the two-dimensional matrix that all thresholds are formed is recorded as Mat_T ₀ ;

(7-3) Perform 3×3 Gaussian filtering on all Mat_T ₀ to output Mat_T ₁ , and then perform image binarization for each region according to Mat_T ₁ .

Preferably, step (8) is specifically:

(8-1) Noise filtering is performed on the image through the morphological opening operation, specifically, an m ₁ ×n ₁ template is constructed first, and then the image is corroded first and then expanded;

(8-2) Merge the vertical edges in the image into connected regions through the morphological closing operation. Specifically, construct an m ₂ ×n ₂ template first, and then perform the operation of first expanding and then corroding the image;

Among them, erosion is to corrode the edge of the object. The specific process is to process the template for each pixel (x, y) in the image as follows: place the pixel (x, y) in the center of the template, and traverse For all other pixels covered by the template, the value of the modified pixel (x, y) is the smallest value among all pixels. Expansion is the opposite of erosion. It is to expand the outline of the object. The specific process is to process each pixel (x, y) in the image as follows: place the pixel (x, y) in the center of the template, and according to the size of the template , traverse all other pixels covered by the template, and modify the value of all pixels in the template to the maximum value among all pixels.

Preferably, step (9) is specifically:

(9-1) Carry out regional labeling to the connected region, and calculate its minimum circumscribed rectangle;

(9-2) According to the aspect ratio ASPECT and the area size AREA of the Chinese license plate, given a deviation rate e ₁ of the aspect ratio and a deviation rate e ₂ of the area, the aspect ratio of the smallest circumscribed rectangle of each connected area is aspect and area perform the following tests:

|aspect-ASPECT|≤e ₁ ASPECT

|area-AREA|≤e ₂ AREA

The connected area that satisfies the above conditions can proceed to step (9-3), otherwise it is a non-license plate area;

(9-3) Generally speaking, the inclination angle of the license plate in the image is not too large, so given a threshold T _angle of the inclination angle, the inclination angle angle of each connected region can pass the following test:

|angle-0|≤T _angle

If the connected area meets the above conditions, it is considered as a license plate area, otherwise it is a non-license plate area.

Compared with the prior art, the present invention has the following advantages and beneficial effects:

1. The present invention uses the Scharr operator instead of the Sobel operator for edge detection, which can detect the vertical edge of the license plate more accurately and maintain the same high efficiency.

2. The present invention innovatively introduces a sharpness evaluation module to make adaptive adjustments to input images with different sharpness, which is more conducive to subsequent processing, and greatly improves the accuracy of license plate positioning under different sharpness.

3. The present invention innovatively introduces a brightness evaluation module to perform illumination normalization processing on input images with abnormal brightness, which is more conducive to subsequent processing and greatly improves the accuracy of license plate positioning in complex illumination environments.

Description of drawings

FIG. 1 is a general flow chart of a method for quickly locating license plates based on clarity and brightness evaluation in the present invention.

detailed description

The present invention will be further described in detail below in conjunction with the embodiments and the accompanying drawings, but the embodiments of the present invention are not limited thereto.

Example

As shown in Figure 1, a kind of license plate quick location method based on sharpness and brightness evaluation in the present embodiment, specifically comprises the following steps:

1. Perform a sharpness evaluation based on noise and sharpness on the input image:

① Noise evaluation is to obtain a binary image by performing OTSU threshold processing on the image first, then calculate the proportion of connected regions whose area is smaller than a given threshold, and finally obtain the noise evaluation value through linear transformation. The formula for calculating the noise evaluation value is as follows:

Among them, Comp _i is the i-th area, Area _i is the area of the i-th area, T _area is a given threshold for judging whether the area is too small, a and b are the parameters of the linear transformation.

②The sharpness evaluation adopts the point sharpness algorithm, and subtracts 8 neighborhood points from each point of the image, and first calculates the weighted sum of the 8 differences (the size of the weight depends on the distance, the distance is long, the weight Small, such as the difference between 45° and 135° direction needs to be multiplied ), add the values obtained from all points and divide by the total number of pixels. The formula for calculating the sharpness evaluation value is as follows:

Among them, a is the pixel in the ath area, m and n are the length and width of the image, df is the change amplitude of the gray level, and dx is the distance increment between pixels.

③ Sharpness evaluation is composed of noise evaluation and sharpness evaluation. After obtaining the noise evaluation value and sharpness evaluation value of the image, the image sharpness evaluation value is obtained through a certain weighted relationship. The formula for calculating the clarity evaluation value is as follows:

Quality=wNoise+(1-w)Sharpness

Among them, w is the weight value, and the value range is between 0 and 1.

2. Perform gradient sharpening on images with insufficient clarity, and perform Gaussian blur processing on images with excessive clarity:

① Two thresholds T _quality1 and T _quality2 are given, where T _quality1 is the threshold for judging whether the image definition is insufficient, and T _quality2 is the threshold for judging whether the image definition is too high.

②When Quality<T _quality1 , the gradient sharpening process is performed on the image, that is, the gradient information Grad of the image is obtained through the Laplace operator, and then added to the original image to obtain the gradient sharpening result. The formula of gradient sharpening is as follows:

f(x,y)=f(x,y)+Grad(x,y)

Among them, f(x, y) is the original image, and Grad(x, y) is the gradient information.

③When Quality>T _quality2 , Gaussian blur processing is performed on the image, that is, the Gaussian function is used to calculate the blur template with a normal distribution, and the Gaussian blur result is obtained by convolution with the original image. The formula of the Gaussian function and Gaussian blur is as follows:

f(x,y)=f(x,y)*Gauss(a,b)

Among them, σ is the standard deviation of the normal distribution, m and n are the length and width of the template, a and b are the elements on the template, and f(x, y) is the original image.

3. Convert from RGB image to grayscale image, 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

Among them, R, G, and B are the values 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. Evaluate the brightness of the grayscale image:

Evaluate the two cases of excessive brightness or insufficient brightness of the grayscale image, specifically calculate the sum of the proportion of pixels whose pixel value is lower than the threshold T _bright1 or higher than the threshold T _bright2 to all pixels, if it is greater than the given threshold T _bright3 It is considered to be an abnormal brightness.

5. If the brightness is abnormal, perform light normalization processing:

①First use Gamma correction to enhance the brightness of the image. The Gamma correction formula is as follows:

g(x,y)=c[f(x,y)] ^γ

In the formula, f(x, y) is the input image, g(x, y) is the output image, c is any value used to adjust the image contrast, and γ is used to adjust the image brightness between 0 and 1;

② Then perform Gaussian difference filtering (DOG, Difference of Gaussian) to remove the high-frequency components of the illumination, that is, the uneven part. The DoG filtering formula is as follows:

g(x,y)=f(x,y)*Gauss(σ ₂ )-f(x,y)*Gauss(σ ₁ )

In the formula, f(x, y) is an input image, g(x, y) is an output image, and Gauss(σ _n ) is a Gaussian function whose variance is σ _n (n=1, 2);

③Finally, contrast equalization is used to enhance the contrast of image illumination, so as to remove unevenness in illumination while retaining useful details of the image to the greatest extent and enhance image contrast. The formula for contrast equalization 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 influence of large pixel values, τ is used for the first Cut off large pixel values in the image after one-step normalization;

6. Extract the vertical edge in the image through the Scharr operator:

In order to extract image edge information, the Sobel operator is commonly used, but because the calculation of the Sobel operator on the 3×3 convolution template is often inaccurate, there is an improved Sobel operator called the Scharr operator, and its operational efficiency As fast as the Sobel operator, but the result is more accurate, the Scharr template expression is as follows:

Since the horizontal edge is generally not beneficial but harmful to the license plate location, only Grad _x is used to convolve the image to obtain its vertical edge. The vertical edge extraction formula is as follows:

f(x,y)=f(x,y)*Grad _x

Among them, f(x, y) is the original image.

7. Perform local adaptive threshold processing:

① Divide the image into m×n small areas of equal size;

② Carry out OTSU processing in each small area, and record the two-dimensional matrix composed of all thresholds as Mat_T ₀ ;

③ Perform a 3×3 Gaussian filter on all Mat_T ₀ to output Mat_T ₁ , and then perform image binarization for each region according to Mat_T ₁ ;

8. Filter noise through morphological processing, and merge vertical edges into connected regions:

① Noise filtering is performed on the image through the morphological opening operation, specifically, an m ₁ ×n ₁ template is constructed first, and then the image is corroded first and then expanded;

② Merge the vertical edges in the image into connected regions through the morphological closing operation. Specifically, construct an m ₂ ×n ₂ template first, and then perform the operation of first dilation and then erosion on the image;

Among them, erosion is to corrode the edge of the object. The specific process is to process the template for each pixel (x, y) in the image as follows: place the pixel (x, y) in the center of the template, and traverse For all other pixels covered by the template, the value of the modified pixel (x, y) is the smallest value among all pixels. Expansion is the opposite of erosion. It is to expand the outline of the object. The specific process is to process each pixel (x, y) in the image as follows: place the pixel (x, y) in the center of the template, and according to the size of the template , traverse all other pixels covered by the template, and modify the value of all pixels in the template to the maximum value among all pixels.

9. Carry out area marking on the connected area, filter according to the characteristics of the Chinese license plate and obtain the license plate area:

①Mark the connected area and calculate its minimum circumscribed rectangle;

②According to the aspect ratio ASPECT and area size AREA of the Chinese license plate, given an aspect ratio deviation rate e ₁ and an area deviation rate e ₂ , make the aspect ratio aspect and area area of the smallest circumscribed rectangle of each connected area Perform the following tests:

|aspect-ASPECT|≤e ₁ ASPECT

|area-AREA|≤e ₂ AREA

Connected areas that meet the above conditions can proceed to step ③, otherwise it is a non-license plate area.

③Generally speaking, the inclination angle of the license plate in the image is not too large, so given a threshold T _angle of inclination angle, the inclination angle angle of each connected region can pass the following test:

|angle-0|≤T _angle

If the connected area meets the above conditions, it is considered as a license plate area, otherwise it is a non-license plate area.

The above-mentioned embodiment is a preferred embodiment of the present invention, but the embodiment of the present invention is not limited by the above-mentioned embodiment, and any other changes, modifications, substitutions, combinations, Simplifications should be equivalent replacement methods, and all are included in the protection scope of the present invention.

Claims (9)

1. a license plate rapid location method based on clarity and brightness evaluation, is characterized in that, comprises the following steps: (1) Carry out sharpness evaluation based on noise and sharpness on the input image; the sharpness evaluation is composed of noise evaluation and sharpness evaluation, after obtaining the noise evaluation value and sharpness evaluation value of the image, the image is obtained through a weighted relationship The clarity evaluation value of , the formula for calculating the clarity evaluation value is as follows: Quality=wNoise+(1-w)Sharpness Among them, w is the weight value, and the value range is between 0 and 1; The noise evaluation is to obtain a binary image by first performing OTSU threshold processing on the image, then calculate the proportion of the area in the connected region smaller than a given threshold, and finally obtain the noise evaluation value through linear transformation. The formula for calculating the noise evaluation value is as follows: <mrow><mi>N</mi><mi>o</mi><mi>i</mi><mi>s</mi><mi>e</mi><mo>=</mo><msub><mi>L</mi><mn>1</mn></msub><munderover><mo>&amp;Sigma;</mo><mrow><msub><mi>Area</mi><mi>i</mi></msub><mo>=</mo><mn>0</mn></mrow><msub><mi>T</mi><mrow><mi>a</mi><mi>r</mi><mi>e</mi><mi>a</mi></mrow></msub></munderover><msub><mi>Comp</mi><mi>i</mi></msub><mo>/</mo><munderover><mo>&amp;Sigma;</mo><mrow><msub><mi>Area</mi><mi>i</mi></msub><mo>=</mo><mn>0</mn></mrow><mi>&amp;infin;</mi></munderover><msub><mi>Comp</mi><mi>i</mi></msub><mo>+</mo><msub><mi>L</mi><mn>2</mn></msub></mrow> Among them, Comp _i is the i-th area, Area _i is the area of the i-th area, T _area is a given threshold for judging whether the area is too small, L ₁ and L ₂ are the parameters of linear transformation; The sharpness evaluation uses the point sharpness algorithm, and subtracts 8 neighborhood points from each point of the image, first calculates the weighted sum of the 8 differences, and then adds and divides the obtained values of all points by the total pixel value. The formula for calculating the sharpness evaluation value is as follows: <mrow><mi>S</mi><mi>h</mi><mi>a</mi><mi>r</mi><mi>p</mi><mi>n</mi><mi>e</mi><mi>s</mi><mi>s</mi><mo>=</mo><munderover><mo>&amp;Sigma;</mo><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mrow><mi>m</mi><mo>&amp;times;</mo><mi>n</mi></mrow></munderover><munderover><mo>&amp;Sigma;</mo><mrow><mi>k</mi><mo>=</mo><mn>1</mn></mrow><mn>8</mn></munderover><mo>|</mo><mfrac><mrow><mi>d</mi><mi>f</mi></mrow><mrow><mi>d</mi><mi>x</mi></mrow></mfrac><mo>|</mo><mo>/</mo><mrow><mo>(</mo><mi>m</mi><mo>&amp;times;</mo><mi>n</mi><mo>)</mo></mrow></mrow> Among them, 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 level, and dx is the distance increment between pixels; (2) Gradient sharpening processing is performed on images with insufficient clarity, and Gaussian blur processing is performed on images with excessive clarity; (3) the image processed in step (2) is converted into a grayscale image by the RGB image; (4) Evaluate the brightness of the grayscale image, that is, evaluate the grayscale image in two situations of excessive brightness or insufficient brightness; (5) If the brightness is abnormal, normalize the illumination; (6) extract the vertical edge in the image by the Scharr operator; (7) Perform local adaptive threshold processing; (8) filter noise through morphological processing, and merge vertical edges into connected regions; (9) Carry out area marking on the connected area, screen according to the characteristics of the Chinese license plate and obtain the license plate area. 2. the license plate rapid location method based on clarity and brightness evaluation according to claim 1, is characterized in that, step (2) is specially: (2-1) Two thresholds T _quality1 and T _quality2 are given, wherein T _quality1 is the threshold for judging whether the image definition is insufficient, and T _quality2 is the threshold for judging whether the image definition is too high; (2-2) When Quality<T _quality1 , the gradient sharpening process is performed on the image, that is, the image gradient information Grad is obtained through the Laplace operator, and then added to the original image to obtain the gradient sharpening result. The gradient sharpening formula is as follows: f1(x,y)=f(x,y)+Grad(x,y) Among them, f(x,y) is the original image, f1(x,y) is the image obtained after sharpening the original image, and Grad(x,y) is the gradient information; (2-3) When Quality>T _quality2 , Gaussian blur processing is performed on the image, that is, the Gaussian function is used to calculate the blur template with a normal distribution, and the Gaussian blur result is obtained by convolution with the original image, Gaussian function and Gaussian blur The formula is as follows: <mrow><mi>G</mi><mi>a</mi><mi>u</mi><mi>s</mi><mi>s</mi><mrow><mo>(</mo><mi>a</mi><mo>,</mo><mi>b</mi><mo>)</mo></mrow><mo>=</mo><mfrac><mn>1</mn><mrow><mn>2</mn><msup><mi>&amp;pi;&amp;sigma;</mi><mn>2</mn></msup></mrow></mfrac><msup><mi>e</mi><mrow><mo>-</mo><mfrac><mrow><msup><mrow><mo>(</mo><mi>a</mi><mo>-</mo><mi>m</mi><mo>/</mo><mn>2</mn><mo>)</mo></mrow><mn>2</mn></msup><mo>+</mo><msup><mrow><mo>(</mo><mi>b</mi><mo>-</mo>mo><mi>n</mi><mo>/</mo><mn>2</mn><mo>)</mo></mrow><mn>2</mn></msup></mrow><mrow><mn>2</mn><msup><mi>&amp;sigma;</mi><mn>2</mn></msup></mrow></mfrac></mrow></msup></mrow> f2(x,y)=f(x,y)*Gauss(a,b) Among them, σ is the standard deviation of the normal distribution, m and n are the length and width of the template, a and b are the elements on the template, f(x,y) is the original image, f2(x,y) is the original image after Gaussian blur the resulting image. 3. the license plate rapid location method based on clarity and brightness evaluation according to claim 1, is characterized in that, in step (3), the formula that changes grayscale image by RGB figure 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; or Among them, R, G, and B are the values 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 fast location method based on clarity and brightness evaluation according to claim 1, is characterized in that, in step (4), the method for carrying out brightness evaluation to grayscale image is: Calculate the sum of the proportions of pixels whose pixel value is lower than the threshold T _bright1 or higher than the threshold T _bright2 to all pixels, and if it is greater than the given threshold T _bright3 , it is considered abnormal brightness. 5. the license plate rapid location method based on clarity and brightness evaluation according to claim 1, is characterized in that, in step (5), the method for illumination normalization processing is: (5-1) First use Gamma correction to enhance the brightness of the image. The Gamma correction formula is as follows: g(x,y)=c[f(x,y)] ^γ In the formula, f(x, y) is the input image, g(x, y) is the output image, c is any value used to adjust the contrast of the image, and γ is used to adjust the brightness of the image between 0 and 1; (5-2) Then perform Gaussian difference filtering to remove the high-frequency component of the illumination, that is, the uneven part. 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 a Gaussian function with variance σ _n (n=1,2); (5-3) Finally, the contrast equalization is used to enhance the contrast of the image illumination, so as to remove the unevenness of the illumination while retaining the useful details of the image to the greatest extent and enhance the image contrast. The contrast equalization formula is as follows: <mrow><msub><mi>g</mi><mn>1</mn></msub><mrow><mo>(</mo><mi>x</mi><mo>,</mo><mi>y</mi><mo>)</mo></mrow><mo>=</mo><mfrac><mrow><mi>f</mi><mrow><mo>(</mo><mi>x</mi><mo>,</mo><mi>y</mi><mo>)</mo></mrow></mrow><msup><mrow><mo>(</mo><mi>m</mi><mi>e</mi><mi>a</mi><mi>n</mi><mo>(</mo><msup><mrow><mo>|</mo><mrow><mi>f</mi><mrow><mo>(</mo><mrow><msup><mi>x</mi><mo>&amp;prime;</mo></msup><mo>,</mo><msup><mi>y</mi><mo>&amp;prime;</mo></msup></mrow><mo>)</mo></mrow></mrow><mo>|</mo></mrow><mi>&amp;alpha;</mi></msup><mo>)</mo><mo>)</mo></mrow><mfrac><mn>1</mn><mi>&amp;alpha;</mi></mfrac></msup></mfrac></mrow> <mrow><msub><mi>g</mi><mn>2</mn></msub><mrow><mo>(</mo><mi>x</mi><mo>,</mo><mi>y</mi><mo>)</mo></mrow><mo>=</mo><mfrac><mrow><msub><mi>g</mi><mn>1</mn></msub><mrow><mo>(</mo><mi>x</mi><mo>,</mo><mi>y</mi><mo>)</mo></mrow></mrow><msup><mrow><mo>(</mo><mi>m</mi><mi>e</mi><mi>a</mi><mi>n</mi><mo>(</mo><mrow><mi>min</mi><msup><mrow><mo>(</mo><mrow><mi>&amp;tau;</mi><mo>,</mo><mrow><mo>|</mo><mrow><msub><mi>g</mi><mn>1</mn></msub><mrow><mo>(</mo><mrow><msup><mi>x</mi><mo>&amp;prime;</mo></msup><mo>,</mo><msup><mi>y</mi><mo>&amp;prime;</mo></msup></mrow><mo>)</mo></mrow></mrow><mo>|</mo></mrow></mrow><mo>)</mo></mrow><mi>&amp;alpha;</mi></msup></mrow><mo>)</mo><mo>)</mo></mrow><mfrac><mn>1</mn><mi>&amp;alpha;</mi></mfrac></msup></mfrac></mrow> <mrow><msub><mi>g</mi><mn>3</mn></msub><mrow><mo>(</mo><mi>x</mi><mo>,</mo><mi>y</mi><mo>)</mo></mrow><mo>=</mo><mi>tanh</mi><mrow><mo>(</mo><mfrac><mrow><msub><mi>g</mi><mn>2</mn></msub><mrow><mo>(</mo><mi>x</mi><mo>,</mo><mi>y</mi><mo>)</mo></mrow></mrow><mi>&amp;tau;</mi></mfrac><mo>)</mo></mrow></mrow> 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 influence of large pixel values, τ is used for the first Cut off large pixel values in the image after one-step normalization. 6. the license plate fast location method based on clarity and brightness evaluation according to claim 1, is characterized in that, in step (6), the concrete method of extracting vertical edge in the image by Scharr operator is: The Scharr template expression is as follows: <mfenced open = "" close = ""><mtable><mtr><mtd><mrow><msub><mi>Grad</mi><mi>x</mi></msub><mo>=</mo><mfenced open = "[" close = "]"><mtable><mtr><mtd><mrow><mo>-</mo><mn>3</mn></mrow></mtd><mtd><mn>0</mn></mtd><mtd><mrow><mo>+</mo><mn>3</mn></mrow></mtd></mtd>mtr><mtr><mtd><mrow><mo>-</mo><mn>10</mn></mrow></mtd><mtd><mn>0</mn></mtd><mtd><mrow><mo>+</mo><mn>10</mn></mrow></mtd></mtr><mtr><mtd><mrow><mo>-</mo><mn>3</mn></mrow></mtd><mtd><mn>0</mn></mtd><mtd><mrow><mo>+</mo><mn>3</mn></mrow></mtd></mtr></mtable></mfenced></mrow></mtd><mtd><mrow><msub><mi>Grad</mi><mi>y</mi></msub><mo>=</mo><mfenced open = "[" close = "]"><mtable><mtr><mtd><mrow><mo>-</mo><mn>3</mn></mrow></mtd><mtd><mrow><mo>-</mo><mn>10</mn></mrow></mtd><mtd><mrow><mo>-</mo><mn>3</mn></mrow></mtd></mtr><mtr><mtd><mn>0</mn></mtd><mtd><mn>0</mn></mtd><mtd><mn>0</mn></mtd></mtr><mtr><mtd><mrow><mo>+</mo><mn>3</mn></mrow></mtd><mtd><mrow><mo>+</mo><mn>10</mn></mrow></mtd><mtd><mrow><mo>+</mo><mn>3</mn></mrow></mtd></mtr></mtable></mfenced></mrow></mtd></mtr></mtable></mfenced> Then use Grad _x to convolve the image to obtain its vertical edge, and the vertical edge extraction formula is as follows: f3(x,y)=f(x,y)*Grad _x Among them, f(x, y) is the original image, and f3(x, y) is the image obtained after the vertical edge extraction of the original image. 7. the license plate fast location method based on clarity and brightness evaluation according to claim 1, is characterized in that, in step (7), the concrete method that carries out local self-adaptive threshold processing is: (7-1) Divide the image into m×n small areas of equal size; (7-2) Carry out OTSU processing in each small area, the two-dimensional matrix that all thresholds are formed is recorded as Mat_T ₀ ; (7-3) Perform 3×3 Gaussian filtering on all Mat_T ₀ to output Mat_T ₁ , and then perform image binarization for each region according to Mat_T ₁ . 8. the license plate quick location method based on clarity and brightness evaluation according to claim 1, is characterized in that, step (8) is specially: (8-1) Noise filtering is performed on the image through the morphological opening operation, specifically, an m ₁ ×n ₁ template is constructed first, and then the image is corroded first and then expanded; (8-2) Merge the vertical edges in the image into connected regions through the morphological closing operation. Specifically, construct an m ₂ ×n ₂ template first, and then perform the operation of first expanding and then corroding the image; Among them, corrosion is to corrode the edge of the object. The specific process is to process the template for each pixel (x, y) in the image as follows: place the pixel (x, y) in the center of the template, and traverse For all other pixels covered by the template, modify the value of the pixel (x, y) to be the smallest value among all pixels. Expansion is the opposite of erosion, which is to expand the outline of the object. The specific process is to apply the template to each pixel in the image (x, y) is processed as follows: place the pixel (x, y) in the center of the template, traverse all other pixels covered by the template according to the size of the template, and modify the value of all pixels in the template to the largest value among all pixels . 9. the license plate quick location method based on clarity and brightness evaluation according to claim 1, is characterized in that, step (9) is specially: (9-1) Carry out regional labeling to the connected region, and calculate its minimum circumscribed rectangle; (9-2) According to the aspect ratio ASPECT and the area size AREA of the Chinese license plate, given a deviation rate e ₁ of the aspect ratio and a deviation rate e ₂ of the area, the aspect ratio of the smallest circumscribed rectangle of each connected area is aspect and area perform the following tests: |aspect-ASPECT|≤e ₁ ASPECT ① |area-AREA|≤e ₂ AREA ② Simultaneously satisfy the connected area of formula ① and ② to carry out step (9-3), otherwise it is a non-license plate area; (9-3) Given a threshold T _angle of an inclination angle, make the inclination angle angle of each connected region pass the following test: |angle-0|≤T _angle ③ If the connected area satisfies the formula ③, it is considered as a license plate area, otherwise it is a non-license plate area.