CN101105893A - Automobile video frequency discrimination speed-testing method - Google Patents
Automobile video frequency discrimination speed-testing method Download PDFInfo
- Publication number
- CN101105893A CN101105893A CNA2006100471946A CN200610047194A CN101105893A CN 101105893 A CN101105893 A CN 101105893A CN A2006100471946 A CNA2006100471946 A CN A2006100471946A CN 200610047194 A CN200610047194 A CN 200610047194A CN 101105893 A CN101105893 A CN 101105893A
- Authority
- CN
- China
- Prior art keywords
- license plate
- image
- character
- vehicle body
- area
- 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
Landscapes
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a vehicle video identification and speed measuring method. The concrete steps are as follows: (a). obtain a vehicle picture; (b). search the partial threshold value of vehicle license plate on the vehicle picture to obtain the license picture; (c). redress the inclination of the license picture; (d). search the frame of the license plate on the redressed picture and calculate the height of characters; (e). standardize characters on the license plate to obtain a number of partial blocks; (f). carry out the binary system for the partial blocks of each character to obtain two-digit pictures; (g). identify characters on the two-digit pictures to obtain characters on the license plate; (h). track the moving route of the license plate through results of step b and d; (i). calculate the moving distance of the vehicle body; and (j). calculate the running speed. The invention has the advantages of high precision of identifying vehicle licenses and low impact on light reflection of license plate, easy realization, and high practical value.
Description
Technical Field
The invention relates to a method for monitoring whether a moving vehicle breaks rules and regulations on a highway, in particular to a vehicle video identification speed measurement method.
Background
With the development of the automobile industry and the continuous improvement of the living standard of people, vehicles become indispensable transportation tools, and a large number of vehicles rush to the street every day, which provides a new subject for traffic management. Whether on a common road or an expressway, illegal vehicles such as overspeed driving and the like can appear, and the traffic safety can not be ensured to the maximum extent. In order to relieve the tension situation, a traffic management department installs and uses an automatic monitoring system on a road, the system identifies the license plate number by carrying out edge extraction, hough transformation, histogram analysis and calculation of morphological operators on the license plate image, the automatic monitoring is suitable for the conditions of certain vehicle speed and stable vehicle flow, but the problem of low identification rate exists for sudden changes, such as escape and passing acceleration of vehicles for avoiding pursuing police cars; the detection method has more operation amount, if voting is carried out by Hough transform, the winning is the area to be determined, although the method is accurate, the method requires larger storage space and calculation time, and is not suitable for real-time application.
Disclosure of Invention
In order to solve the problems that the prior art has low recognition rate and is not suitable for real-time application to the fast passing vehicle, the invention aims to provide a vehicle video recognition speed measurement method capable of real-time detection no matter in daytime or at night.
In order to achieve the purpose, the invention adopts the technical scheme that:
the method comprises the following steps:
a. acquiring a vehicle body image;
b. searching a local threshold value of the license plate on the vehicle body image to obtain a license plate image;
c. correcting the inclination angle of the license plate image;
d. searching a license plate frame area on the license plate image after the inclination angle correction, and calculating the height of the characters;
e. normalizing license plate character patterns in the license plate image to obtain a plurality of local small blocks;
f. carrying out binarization on the local small blocks of each character and outputting a binary image;
g. carrying out character recognition on the binary image to obtain license plate number characters;
h. following the step d, tracking the moving path of the license plate in a group of continuous images by using the results of the step b and the step d;
i. calculating the moving distance of the vehicle body;
j. and calculating the running speed of the vehicle.
The step b of searching the local threshold value of the license plate on the vehicle body image further comprises the following steps:
b1. dividing a license plate area and a background area;
b2. performing edge detection on all license plate images, and dividing the license plate images into small blocks with the same size according to unit pixels (8*8 is taken as a unit in the embodiment);
b3. calculating a reference value for each small block, and taking one small block as a unit for all license plate images to obtain a local row and column reference value of the license plate images;
b4. and binarizing the license plate image according to the row and column reference value to obtain a binary image of the license plate.
The step c of correcting the inclined angle of the license plate image comprises the following steps:
c1. segmenting a character area in the license plate area;
c2. calculating the line direction dispersion sum in the character area;
c3. solving the line direction dispersion sum in all the number plate areas;
c4. calculating the line direction dispersion density of the non-character area on all license plates;
c5. repeating the processes of c1 to c4 in the license plate rotating image between the degrees of minus 5 degrees and 5 degrees, and solving the line direction dispersion density of the non-character area in each rotating image;
c6. when the value d is minimum, the final inclination angle of the license plate is the rotation angle of the license plate to be corrected.
Step d, searching a license plate frame area, and calculating the word height comprises the following steps:
d1. strengthening the license plate image;
d2. separating each character area by using the specification information of the license plate, determining the character position, namely searching the minimum area of the blank area brightness sum on the license plate image to determine the blank area between the characters on the premise of meeting the license plate proportion, and further determining the character area;
d3. obtaining the character height according to the standard specification information of the license plate;
the step e of standardizing the license plate number character patterns specifically comprises the following steps: the text size is normalized to 16 by 16 local tiles.
Said step f binarizes the local small block of each character, and outputting a binary image further includes:
f1. dividing the local small block obtained in the step e into a plurality of square blocks;
f2. acquiring 4 Fang Kuaizu with twice side length of the same square block;
f3. calculating all area reference values in 4 square block groups with double side lengths;
f4. determining the reference value of the square block by using the average value of the reference values of all the regions;
f5. on the character image of the local small block, the area reference value of each other block is calculated by the same method;
f6. obtaining a threshold matrix of the character image of the local small block according to the result;
f7. and carrying out 2-system on each character image according to the threshold matrix to obtain a binary image.
And g, recognizing the characters of the binary image obtained in the step f by using an artificial neural network method.
The step h of determining the license plate moving path in a group of continuous images comprises the following steps:
h1. recording the time difference between a current frame and a previous frame and the time difference between the current frame and a later frame in an image column in real time, and acquiring two edge images;
h2. respectively binarizing the two acquired edge images to obtain two binary images;
h3. performing logic operation on the two binary images;
h4. judging the size of the external figure of the vehicle body by a histogram method in the operation result;
h5. when the vehicle body displays the maximum external graph on the image, the ratio of the height pixel of the license plate based on the whole vehicle body image to the height pixel of the whole vehicle body image is calculated, and the ratio is the 1 st ratio;
h6. calculating the ratio of the width pixel of the starting position of the license plate on the whole vehicle body image to the width pixel of the whole vehicle body image, wherein the ratio is the 2 nd ratio;
h7. calculating the ratio of height (or width) pixels of the vehicle body image just after entering the visual field of the camera to height (or width) pixels of the vehicle body image before exiting the visual field of the camera, wherein the ratio is a 3 rd ratio;
h8. calculating the coordinates of the license plate in the whole image at the upper left corner through the ratio of 1 st to 3 rd;
h9. and acquiring the central coordinates of the license plate on all the frames, and determining the motion track of the license plate on all the image columns.
The step i of calculating the moving distance of the vehicle body specifically comprises the following steps: when the free focal length lens of the camera device is fixed, the distance between the camera device and the license plate is obtained according to the number of pixels in the length direction of the license plate.
The invention has the following beneficial effects and advantages:
1. the accuracy rate of identifying the license plate number is high. The method utilizes image signal processing technologies such as inclination angle correction and character region segmentation of the license plate image and a mode recognition technology of an artificial neural network, and compared with a Hough conversion method, improves the license plate recognition rate and the accuracy rate of vehicle speed detection, wherein the license plate recognition rate is 99.95% in the daytime, 98% at night, and the vehicle speed detection error range is +/-3 Km/h;
2. the influence on the light reflection of the license plate is reduced. The invention adopts a local binarization method for the license plate image, thereby effectively solving the problem that character recognition is influenced by light reflection of the light license plate;
3. is suitable for real-time processing. The invention effectively utilizes the information resources of the vehicle equipment, such as the specification characteristics of the license plate, to identify the license plate number, so the identification speed is very high, the moving track of the vehicle is searched in real time and the characters of the license plate are identified, the overspeed vehicle can be timely monitored, the vehicle speed identification processing time is within 200ms, and the violation vehicle can be rapidly captured and determined through the license plate number identification;
4. easy to realize and strong in practicability. Because the method only utilizes the image signal to carry out identification or detection, the device applying the method does not need an infrared transceiver, does not need to be set in the daytime or at night, and can accurately identify the characters of the license plate even when the vehicle is on and drives from the head to the head.
Drawings
FIG. 1 is a block diagram of the present invention;
FIGS. 2 (a) - (e) are the resulting images and data outputted during the image processing process of the present invention
FIG. 3 is a schematic diagram of license plate specification information applied in the method of the present invention;
FIGS. 4 (a) - (e) are schematic diagrams of the process of standardizing the license plate number text patterns in the method of the present invention;
fig. 5 is a schematic view of calculating an actual moving distance of the vehicle body.
Detailed Description
The method of the present invention is implemented by a control program installed in a hardware device, as shown in figure 1,
the method comprises the following steps:
a. acquiring a vehicle body image;
b. searching a local threshold value of the license plate on the vehicle body image to obtain a license plate image;
c. correcting the inclination angle of the license plate image;
d. searching a license plate frame area on the license plate image after the inclination angle correction, and calculating the character height;
e. standardizing the pattern of license plate characters in the license plate image, and dividing local small blocks on the basis of each character;
f. binarizing the local small blocks of each character and outputting binary images;
g. carrying out character recognition on the binary image to obtain license plate number characters;
h. c, following the step d, tracking the moving path of the license plate in a group of continuous images by using the results of the step b and the step d;
i. calculating the actual moving distance of the vehicle body according to the moving path of the license plate;
j. and calculating the running speed of the vehicle.
The step b of searching the local threshold value of the license plate on the vehicle body image further comprises the following steps:
b1. dividing a license plate region and a background region by using an extremum solving method, wherein the license plate region and the background region are divided by using a minimum value point solving method in the embodiment;
b2. performing edge detection on all license plate images, and dividing the license plate images into small blocks with the same size according to unit pixels (8*8 is taken as a unit in the embodiment);
b3. calculating a brightness reference value (threshold value) for each small block, and obtaining a local row-column reference value (local brightness threshold value matrix) of the license plate image by taking one small block as a unit for all license plate images, namely calculating the brightness reference value for each block 8*8 for 1 time:
T1 i，j ＝M i，j +2.26 i，j (1)
wherein M is i，j Average value of brightness of all license plate images, 6 i，j Standard deviation of brightness of 8 × 8block; since each 8*8 small block repeats the above process, a localized threshold matrix (Thresholdmatrix) can be obtained for the car body image by using 8*8 as a unit;
b4. the method comprises the following steps of binarizing a license plate image according to the Threshold matrix (Threshold matrix) to obtain a binary image of the license plate, namely binarizing all license plate images according to the Threshold matrix, wherein the occurrence rate of a license plate area 1 on the binary image is high, the blank characteristic of the license plate can be correctly reflected, and the license plate field can be easily detected by utilizing the specification characteristic of the license plate, and the specific method comprises the following steps:
performing morphological (morphological) operation of the obtained binary image with unit matrix (element matrix) size of (3,3) and operation form of open (exposure); then, performing morphological (morphology) operation with unit matrix (element matrix) of (16,3) size to calculate the morphological (dilation) as expansion; the number plate area is obtained from the finally acquired image by using the characteristics of the number plate (the size of the lateral and vertical directions, the ratio of the lateral and vertical directions, and the black and white characteristics of the number plate).
As shown in fig. 2 (a), the image is a license plate region image obtained by dividing a license plate region and a background region.
The step c of correcting the inclined angle of the license plate image comprises the following steps:
c1. segmenting a character area in the license plate area;
on the acquired license plate image, characters need to be subjected to region segmentation, and due to the fact that the license plate is inclined frequently, the recognition rate is difficult to reduce. Therefore, before the character is divided into the sections, the inclination angle of the inclined license plate needs to be corrected. The maximum inclination angle in the present embodiment is limited to [ -5 °, +5 ° ]; the basic idea is that when the license plate is upright and horizontally placed, the line direction dispersion density is minimum on the blank areas of the upper part and the lower part of the license plate; the line direction dispersion density is obtained by summing dispersion Values (VARIANCE) for each line unit and dividing the sum by the number of lines.
Differentiating in the row direction by taking the pixel width of the font thickness as a unit to obtain a row direction differential matrix; accumulating the differential matrix in the row direction to obtain a differential sum S i (summatontono differrnce) that contains column vector (columnirection) information;
S i ＝∑ N j＝1 (P ij -P ij+a )
wherein: s i Is the differential sum of i rows; n is the image width (in pixels); a is the minimum unit pixel of the font width; p ij Is the luminance value of the (i, j) pixel; p ij+a Is the luminance value of the (i, j + a) pixel.
And calculating the brightness average value of the column vector, and when the brightness average value of the column vector is continuously larger than the average value, considering the area corresponding to the column vector as a character area.
c2. Calculating the line DIRECTION dispersion sum (ROW DIRECTION sum OF VARIANCE sum OF line DIRECTIONs) in the character area;
Vi＝1/N(∑ N j＝1 (M i -P ij ) 2 ) 1/2
M i ＝1/N∑ N j＝1 (P ij /N)
V 0 ＝∑ H0 i＝1 V i
wherein: h0 is the height (in pixels) of the text bar region; vi is the variance value (variance value) of the ith row; mi is the average brightness value of the ith row; v 0 Is the sum of the dispersion values in row i;
c3. calculating the line direction dispersion sum in all license plate areas;
Vi＝1/N(∑ N j＝1 (M i -P ij ) 2 ) 1/2
M i ＝1/N∑ N j＝1 (P ij /N)
V＝∑ H i＝1 V i
wherein: v is the sum of the dispersion values in all the areas of the license plate; h is the total height of the license plate.
c4. The line direction dispersion density of the non-text area was calculated on all license plates:
d＝(V-V 0 )/(H-H 0 )
wherein: d is the line direction dispersion density of the non-character area;
c5. repeating the processes of c1 to c4 in the license plate rotating image between the degrees of minus 5 degrees and 5 degrees, and solving the line direction dispersion density of the non-character area in each rotating image;
c6. when the value d is minimum, the final inclination angle of the license plate is the rotation angle of the license plate to be corrected:
[d，α]＝min(d(K)；
wherein: alpha is the inclination angle of the license plate, the dereferencing range is [ -5 degrees, +5 degrees ], when d (k) is the minimum rotation angle, namely the inclination angle of the license plate to be corrected;
d (K) is the line direction dispersion density of the non-character area in the K-th rotation image;
as shown in fig. 2 (b), the license plate image is corrected by the tilt angle and divided into character regions.
Step d, searching a license plate frame area, and calculating the word height comprises the following steps:
d1. strengthening the license plate image;
d2. dividing each character area by using the specification information of the license plate, determining the character position, namely determining a blank area between characters by searching a minimum value area of the brightness sum on the license plate image on the premise of meeting the proportion characteristic of the license plate, and further determining the character area;
the character segmentation utilizes the specification information of the license plate, only one scanning FIELD (FIELD, which is equivalent to a half frame) of each frame is utilized on a dynamic image column, so that the license plate image is reduced by half in the vertical direction compared with the actual license plate specification. Only one scanning field of each frame (frame) is utilized on the dynamic image array, so that the vertical distance of the dynamic image array is reduced by half compared with the vertical distance of an actual license plate, the input license plate image is reduced by half compared with the side length ratio (side ratio) of the actual license plate specification, the side length ratio of the actual license plate is 440: 130, and the side length ratio of the license plate on the license plate image is 440: 65.
As shown in fig. 3, the relationship between the height of the characters on the license plate and the length of the license plate is as follows:
L＝(9+34/45)H0
wherein: l is the length of the license plate; h0 is the height of characters on the license plate;
converting the above-mentioned character height H0 from 0.8-1.2 times, determining the license plate specification size of every character height, and utilizing said conversion to reduce error of image data;
the pixel points corresponding to the black portion in fig. 3 are used as comparison points between the license plate specification and the actual image.
And searching the position where the brightness sum on the blank area of the license plate standard specification is minimum on the license plate image.
Taking the lower left point of the license plate as a coordinate origin, taking two sides of the license plate as two coordinate axes to establish a rectangular coordinate system, taking the distance from the coordinate origin to a search starting point as P0, and the search height Ha = a × H0, wherein a is a value in the range of 0.8-1.2, and a comparison point used for searching the standard specification of the license plate, namely a black area in an image is determined according to the following mode:
m1＝60/45*a*H0； n1＝72/45*a*H0；
m2＝117/45*a*H0；n2＝129/45*a*H0；
m3＝174/45*a*H0；n3＝208/45*a*H0；
m4＝253/45*a*H0；n4＝265/45*a*H0；
m5＝310/45*a*H0；n5＝322/45*a*H0；
m6＝367/45*a*H0；n6＝379/45*a*H0；
wherein: m 1-m 6 are respectively the initial position horizontal coordinates of the 1 st to 6 th black areas in the license plate image in the coordinate system; n 1-n 6 are respectively the abscissa of the ending position of the 1 st to 6 th black areas in the license plate image in the coordinate system.
The sum of the brightness of the pixels in the blank area of the license plate image standard specification is S,
Sa，p 0 ＝∑ 6 k＝1 ∑ a*H0 i＝0 ∑ nk j＝mk P ij
the corresponding values of a and P0 are obtained when the luminance sum S is minimum:
S min ＝min 0.8＜a＜1.2 {min 0＜p0＜Δp {S a.p0 }}
wherein: delta p is the sum of the step lengths converted after each scanning on the license plate image, and one step length in the embodiment is 1 image;
and determining the starting position of the license plate in the license plate image according to the P0 value at the moment, determining a proper character height range according to the a value at the moment, finally determining the corresponding character height HO when the sum of the brightness on the blank area of the license plate standard specification is minimum, and determining the character positions of each license plate in the license plate image according to the license plate standard specification. As shown in fig. 2 (c).
d3. Then according to the standard specification information (proportion characteristic) of the license plate, finally obtaining the character height pixels on the license plate image;
the step e of normalizing the license plate number character pattern comprises the following steps:
e1. defining the text size as 16 × 16 local small blocks, as shown in fig. 4 (a); this embodiment employs an INTERPOLATION (INTERPOLATION).
Said step f binarizes the 16 × 16 local small blocks of each character, and outputs a binary image;
f1. decomposing the character image of the 16 × 16 local small blocks obtained in step e into 4*4 squares as shown in fig. 4 (a);
f2. obtaining 4 Fang Kuaizu with two times of the side length of the same 4*4 square BLOCK, namely 4 8*8 BLOCKs (BLOCK), as shown in fig. 4 (b) - (e);
f3. all region reference values (threshold values) in 4 8*8 blocks were obtained by the OTSU method
f4. Determining the reference value of 4*4 square by using the average value of the reference values of all the areas;
f5. calculating the area reference value of each 4*4 square on the 16-by-16 character image by the same method;
f6. obtaining a threshold matrix about the 16-by-16 character image according to the result;
f7. each text image is binarized according to the threshold matrix.
Performing character recognition on the binary image obtained in the step f in the step g to obtain the license plate number characters, as shown in fig. 2 (d); the invention utilizes 4 layers of artificial neural networks for identifying English characters and numbers and 3 layers of artificial neural networks for identifying Chinese characters, wherein the 4 layers of artificial neural networks are 256 in number of inlet layers, 160 in number of nerve cells (neuron) of a hidden layer 1, 80 in number of nerve cells (neuron) of a hidden layer 2 and 34 in number of outlet layers. The update of the connection strength and the threshold value, using a gradient-gradient method (gradient-gradient-gradient method)
In order to improve the recognition rate and induction (generation) capability of the artificial neural network, the number of learning modes (LEARN PATTERN) of each character is consistent, and an artificial mode image and a statistical noise image are added.
The step h of determining the license plate moving path in a group of continuous images comprises the following steps:
h1. recording the time difference between the current frame and the previous frame and the time difference between the current frame and the next frame in the image column in real timeExample two edge images E are acquired using the PREWITTS operator 1 、E 2 ：
E 1 ＝Prewitt(F b -F c )
E 2 ＝Prewitt(F f -F c )
Wherein: f c Current FRAME (FRAME); f b Is the previous frame; f f For a later frame;
h2. the two acquired edge images are respectively binarized to obtain two binary images Th1 and Th2:
Th1＝Threshold(E1)；
Th2＝Threshold(E2)；
the binarization process utilizes the standard values for all thresholds for all images as follows:
T＝M+36；
wherein: m is the average brightness value of the edge images E1 and E2; 6 is the standard deviation of the brightness of the edge images E1 and E2;
h3. performing AND (AND) operation on the two binary images to obtain a circumscribed graph (a circumscribed rectangle in the embodiment) of the vehicle body image;
Th＝(Th1)AND(Th2)；
h4. judging the size of the circumscribed graph of the vehicle body image in the circumscribed rectangle by a histogram method;
h5. when the vehicle body displays the maximum external graph on the image, the ratio of the height pixel of the license plate based on the whole vehicle body image to the height pixel of the whole vehicle body image is calculated, and the ratio is the 1 st ratio;
h6. calculating the ratio of the width pixel of the starting position of the license plate on the whole vehicle body image to the width pixel of the whole vehicle body image, wherein the ratio is the 2 nd ratio;
h7. calculating the ratio of height (or width) pixels of the vehicle body image just after entering the visual field of the camera to height (or width) pixels of the vehicle body image before exiting the visual field of the camera, wherein the ratio is a 3 rd ratio;
h8. calculating the coordinates of the license plate in the whole image at the upper left corner through the ratios of 1 st to 3 rd;
h9. and obtaining the central coordinates of the license plate on all the frames, and determining the motion track of the license plate on all the image columns.
The step j of calculating the moving distance of the vehicle body further comprises:
i1. when a free focal length lens of the camera device is fixed, the distance between the camera device and a license plate is solved according to the number of pixels in the length direction of the license plate;
in order to determine the vehicle speed, the distance the vehicle has to travel within a certain period of time must be known; the distance traveled by the vehicle in the current image, the camera free focal length lens state, is first determined, and then the corresponding distance in real space per pixel is known.
In order to test the distance, the length of the license plate is utilized. In the state that the free focal length lens of the camera is fixed, the distance between the camera and the license plate number is calculated according to the number of pixels in the length direction of the license plate,
as shown in fig. 5, if the camera is installed at a point O ', its ground installation height OO ' is h, the installation angle of the camera is α, the viewing angle is α ', the center coordinate of the license plate in the vehicle body image is P, the point C ' is an arbitrary position where the vehicle enters the field of view of the camera, and the distance OC ' from the point O is y:
wherein: -w/2 < p < w/2,w is the number of height pixels on the display screen;
j2. calculating vehicle speed
In the vehicle body image, the actual distance corresponding to the pixels p1 (i 1, j 1) -p2 (i 2, j 2) where the vehicle moves is calculated according to the following formula:
V＝Y/t
Y＝abs(y(i1)-y(i2))
t = N/frame/second
The time t is determined by the number of frames N that the pixels of the vehicle move from p1 to p2, V is the vehicle speed, and Y is the distance actually traveled by the vehicle during the time t, and as shown in fig. 2 (e), is the vehicle speed and the recognition time output in the present embodiment.
Claims (9)
1. A vehicle video identification speed measurement method is characterized by comprising the following steps:
a. acquiring a vehicle body image;
b. searching a local threshold value of the license plate on the vehicle body image to obtain a license plate image;
c. correcting the inclination angle of the license plate image;
d. searching a license plate frame area on the license plate image after the inclination angle correction, and calculating the height of the characters;
e. normalizing license plate character patterns in the license plate image to obtain a plurality of local small blocks;
f. binarizing the local small blocks of each character and outputting binary images;
g. carrying out character recognition on the binary image to obtain license plate number characters;
h. following the step d, tracking the moving path of the license plate in a group of continuous images by using the results of the step b and the step d;
i. calculating the moving distance of the vehicle body;
j. and calculating the running speed of the vehicle.
2. The vehicle video identification speed measurement method according to claim 1, characterized in that: the step b of searching the local threshold value of the license plate on the vehicle body image further comprises the following steps:
b1. dividing a license plate area and a background area;
b2. carrying out edge detection on all license plate images, and dividing the license plate images into small blocks with the same size according to unit pixels;
b3. calculating a reference value for each small block, and taking one small block as a unit for all license plate images to obtain a local row and column reference value of the license plate images;
b4. and binarizing the license plate image according to the row and column reference value to obtain a binary image of the license plate.
3. The vehicle video identification speed measurement method according to claim 1, characterized in that: the step c of correcting the inclined angle of the license plate image comprises the following steps:
c1. segmenting a character area in the license plate area;
c2. obtaining the line direction dispersion sum in the character area;
c3. solving the line direction dispersion sum in all the number plate areas;
c4. calculating the line direction dispersion density of the non-character area on all license plates;
c5. repeating the processes of c1 to c4 in the license plate rotating image between the degrees of minus 5 degrees and 5 degrees, and solving the line direction dispersion density of the non-character area in each rotating image;
c6. when the value d is minimum, the final inclination angle of the license plate is the rotation angle of the license plate to be corrected.
4. The vehicle video identification speed measurement method according to claim 1, characterized in that: step d, searching a license plate frame area, and calculating the word height comprises the following steps:
d1. strengthening the license plate image;
d2. separating each character area by using the specification information of the license plate, determining the character position, namely searching a minimum area of the blank area brightness sum on the license plate image to determine a blank area between characters on the premise of meeting the license plate proportion, and further determining a character area;
d3. and obtaining the character height according to the standard specification information of the license plate.
5. The vehicle video identification speed measurement method according to claim 1, characterized in that: the step e of standardizing the license plate number character patterns specifically comprises the following steps: the text size is normalized to 16 by 16 local tiles.
6. The vehicle video identification speed measurement method according to claim 1, characterized in that: said step f binarizes the local small block of each character, and outputting a binary image further includes:
f1. dividing the local small block obtained in the step e into a plurality of square blocks;
f2. acquiring 4 blocks of Fang Kuaizu with two times of side length of the same block;
f3. calculating all area reference values in the block groups with 4 double side lengths;
f4. determining the reference value of the square block by using the average value of the reference values of all the regions;
f5. on the character image of the local small block, the area reference value of each other block is solved by the same method;
f6. obtaining a threshold matrix of the character image of the local small block according to the result;
f7. and carrying out 2-system on each character image according to the threshold matrix to obtain a binary image.
7. The vehicle video identification speed measurement method according to claim 1, characterized in that: and g, carrying out character recognition on the binary image obtained in the step f by using an artificial neural network method.
8. The vehicle video identification speed measurement method according to claim 1, characterized in that: the step h of determining the license plate moving path in a group of continuous images comprises the following steps:
h1. recording the time difference between the current frame and the previous frame and the time difference between the current frame and the next frame in the image column in real time, and acquiring two edge images;
h2. the two acquired edge images are respectively binarized to obtain two binary images;
h3. performing logic operation on the two binary images;
h4. judging the size of the external figure of the vehicle body by a histogram method in the operation result;
h5. when the vehicle body displays the maximum external graph on the image, the ratio of the height pixel of the license plate based on the whole vehicle body image to the height pixel of the whole vehicle body image is calculated, and the ratio is the 1 st ratio;
h6. calculating the ratio of the width pixel of the starting position of the license plate on the whole vehicle body image to the width pixel of the whole vehicle body image, wherein the ratio is the 2 nd ratio;
h7. calculating the ratio of the height or width pixel of the vehicle body image just after entering the camera view field to the height or width pixel of the vehicle body image before exiting the camera view field, wherein the ratio is a 3 rd ratio;
h8. calculating the coordinates of the license plate in the whole image at the upper left corner through the ratio of 1 st to 3 rd;
h9. and obtaining the central coordinates of the license plate on all the frames, and determining the motion track of the license plate on all the image columns.
9. The vehicle video identification speed measurement method according to claim 1, wherein the step i of calculating the vehicle body movement distance specifically comprises: when the free focal length lens of the camera device is fixed, the distance between the camera device and the license plate is obtained according to the number of pixels in the length direction of the license plate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100471946A CN100452110C (en) | 2006-07-14 | 2006-07-14 | Automobile video frequency discrimination speed-testing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100471946A CN100452110C (en) | 2006-07-14 | 2006-07-14 | Automobile video frequency discrimination speed-testing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101105893A true CN101105893A (en) | 2008-01-16 |
| CN100452110C CN100452110C (en) | 2009-01-14 |
Family
ID=38999776
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2006100471946A Expired - Fee Related CN100452110C (en) | 2006-07-14 | 2006-07-14 | Automobile video frequency discrimination speed-testing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100452110C (en) |
Cited By (25)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101833863A (en) * | 2010-05-17 | 2010-09-15 | 瑞斯康达科技发展股份有限公司 | Method and device for detecting vehicle flow speed, as well as method and system for controlling traffic lights |
| CN101937614A (en) * | 2010-06-12 | 2011-01-05 | 北京中科卓视科技有限责任公司 | Plug and play comprehensive traffic detection system |
| CN101976340A (en) * | 2010-10-13 | 2011-02-16 | 重庆大学 | License plate positioning method based on compressed domain |
| GB2472793A (en) * | 2009-08-17 | 2011-02-23 | Pips Technology Ltd | Measuring the speed of a vehicle using the relative size of a feature of the vehicle in an image to determine distance to the vehicle. |
| CN102147858A (en) * | 2011-03-31 | 2011-08-10 | 重庆大学 | License plate character identification method |
| CN102156480A (en) * | 2010-12-30 | 2011-08-17 | 清华大学 | Unmanned helicopter independent landing method based on natural landmark and vision navigation |
| CN102332209A (en) * | 2011-02-28 | 2012-01-25 | 王志清 | Automobile violation video monitoring method |
| CN102339531A (en) * | 2010-07-14 | 2012-02-01 | 数伦计算机技术（上海）有限公司 | Road traffic detection system |
| CN102436744A (en) * | 2011-09-20 | 2012-05-02 | 中盟智能科技（苏州）有限公司 | Method and device for locking license plate picture |
| US20130229517A1 (en) * | 2012-03-02 | 2013-09-05 | Xerox Corporation | Vehicle speed measurement method and system utilizing a single image capturing unit |
| CN103605960A (en) * | 2013-11-15 | 2014-02-26 | 长安大学 | Traffic state identification method based on fusion of video images with different focal lengths |
| EA020239B1 (en) * | 2010-07-20 | 2014-09-30 | Общество С Ограниченной Ответственностью "Технологии Распознавания" | Method for determining the speed of a vehicle |
| CN104318782A (en) * | 2014-10-31 | 2015-01-28 | 浙江力石科技股份有限公司 | Expressway video speed measuring method and system for zone overlapping |
| WO2015147764A1 (en) * | 2014-03-28 | 2015-10-01 | Kisa Mustafa | A method for vehicle recognition, measurement of relative speed and distance with a single camera |
| CN107292932A (en) * | 2016-04-07 | 2017-10-24 | 上海交通大学 | Head-on video frequency speed-measuring method based on image expansion rate |
| CN107392313A (en) * | 2017-06-12 | 2017-11-24 | 五邑大学 | A kind of rail recognition methods based on deep learning |
| CN107730614A (en) * | 2017-01-05 | 2018-02-23 | 西安艾润物联网技术服务有限责任公司 | Parking charge method and device |
| CN108052866A (en) * | 2017-11-17 | 2018-05-18 | 克立司帝控制系统（上海）有限公司 | Car license recognition learning method and system based on artificial neural network |
| CN108320531A (en) * | 2018-04-04 | 2018-07-24 | 武汉市技领科技有限公司 | A kind of speed measuring equipment and velocity-measuring system |
| CN109615879A (en) * | 2018-12-28 | 2019-04-12 | 成都路行通信息技术有限公司 | Speed abnormity early warning model, method and model building method based on car networking |
| CN109767629A (en) * | 2019-02-01 | 2019-05-17 | 深圳市润泽材料科技有限公司 | A kind of camera system and its speed-measuring method of integrated speed measuring function |
| CN110632339A (en) * | 2019-10-09 | 2019-12-31 | 天津天地伟业信息系统集成有限公司 | Water flow testing method of video flow velocity tester |
| CN110705493A (en) * | 2019-10-10 | 2020-01-17 | 深圳市元征科技股份有限公司 | Method and system for detecting vehicle running environment, electronic device and storage medium |
| CN112199545A (en) * | 2020-11-23 | 2021-01-08 | 湖南蚁坊软件股份有限公司 | Keyword display method and device based on picture character positioning and storage medium |
| CN113030506A (en) * | 2021-03-25 | 2021-06-25 | 上海其高电子科技有限公司 | Micro-area speed measurement method and system based on multi-license plate calibration library |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0668376A (en) * | 1992-08-17 | 1994-03-11 | Omron Corp | Speed violation warning device |
| JPH0883393A (en) * | 1994-09-14 | 1996-03-26 | Hitachi Ltd | Vehicle speed measuring instrument |
| EP1056064B1 (en) * | 1999-05-28 | 2007-12-12 | Nippon Telegraph and Telephone Corporation | Apparatus and method for speed measurement of vehicles with an image processing system |
| JP2001126184A (en) * | 1999-10-29 | 2001-05-11 | Matsushita Electric Ind Co Ltd | Automatic license plate recognizing device and vehicle speed measuring method |
| JP2003228793A (en) * | 2002-02-04 | 2003-08-15 | Matsushita Electric Ind Co Ltd | Reader for front and rear license plates |
| CN1737578A (en) * | 2004-08-19 | 2006-02-22 | 昆明利普机器视觉工程有限公司 | Road vehicle speed measuring method realized only by video |
-
2006
- 2006-07-14 CN CNB2006100471946A patent/CN100452110C/en not_active Expired - Fee Related
Cited By (37)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2472793A (en) * | 2009-08-17 | 2011-02-23 | Pips Technology Ltd | Measuring the speed of a vehicle using the relative size of a feature of the vehicle in an image to determine distance to the vehicle. |
| US8964031B2 (en) | 2009-08-17 | 2015-02-24 | 3M Innovative Properties Company | Method and system for measuring the speed of a vehicle |
| GB2472793B (en) * | 2009-08-17 | 2012-05-09 | Pips Technology Ltd | A method and system for measuring the speed of a vehicle |
| CN101833863A (en) * | 2010-05-17 | 2010-09-15 | 瑞斯康达科技发展股份有限公司 | Method and device for detecting vehicle flow speed, as well as method and system for controlling traffic lights |
| CN101833863B (en) * | 2010-05-17 | 2013-09-11 | 瑞斯康达科技发展股份有限公司 | Method and device for detecting vehicle flow speed, as well as method and system for controlling traffic lights |
| CN101937614A (en) * | 2010-06-12 | 2011-01-05 | 北京中科卓视科技有限责任公司 | Plug and play comprehensive traffic detection system |
| CN102339531B (en) * | 2010-07-14 | 2014-11-05 | 数伦计算机技术（上海）有限公司 | Road traffic detection system |
| CN102339531A (en) * | 2010-07-14 | 2012-02-01 | 数伦计算机技术（上海）有限公司 | Road traffic detection system |
| EA020239B1 (en) * | 2010-07-20 | 2014-09-30 | Общество С Ограниченной Ответственностью "Технологии Распознавания" | Method for determining the speed of a vehicle |
| CN101976340B (en) * | 2010-10-13 | 2013-04-24 | 重庆大学 | License plate positioning method based on compressed domain |
| CN101976340A (en) * | 2010-10-13 | 2011-02-16 | 重庆大学 | License plate positioning method based on compressed domain |
| CN102156480A (en) * | 2010-12-30 | 2011-08-17 | 清华大学 | Unmanned helicopter independent landing method based on natural landmark and vision navigation |
| CN102332209A (en) * | 2011-02-28 | 2012-01-25 | 王志清 | Automobile violation video monitoring method |
| CN102147858B (en) * | 2011-03-31 | 2012-08-01 | 重庆大学 | License plate character identification method |
| CN102147858A (en) * | 2011-03-31 | 2011-08-10 | 重庆大学 | License plate character identification method |
| CN102436744A (en) * | 2011-09-20 | 2012-05-02 | 中盟智能科技（苏州）有限公司 | Method and device for locking license plate picture |
| CN102436744B (en) * | 2011-09-20 | 2014-12-10 | 中盟智能科技（苏州）有限公司 | Method and device for locking license plate picture |
| US20130229517A1 (en) * | 2012-03-02 | 2013-09-05 | Xerox Corporation | Vehicle speed measurement method and system utilizing a single image capturing unit |
| US9083856B2 (en) * | 2012-03-02 | 2015-07-14 | Xerox Corporation | Vehicle speed measurement method and system utilizing a single image capturing unit |
| CN103605960A (en) * | 2013-11-15 | 2014-02-26 | 长安大学 | Traffic state identification method based on fusion of video images with different focal lengths |
| WO2015147764A1 (en) * | 2014-03-28 | 2015-10-01 | Kisa Mustafa | A method for vehicle recognition, measurement of relative speed and distance with a single camera |
| CN104318782A (en) * | 2014-10-31 | 2015-01-28 | 浙江力石科技股份有限公司 | Expressway video speed measuring method and system for zone overlapping |
| CN107292932B (en) * | 2016-04-07 | 2020-06-09 | 上海交通大学 | Head-on video speed measurement method based on image expansion rate |
| CN107292932A (en) * | 2016-04-07 | 2017-10-24 | 上海交通大学 | Head-on video frequency speed-measuring method based on image expansion rate |
| CN107730614A (en) * | 2017-01-05 | 2018-02-23 | 西安艾润物联网技术服务有限责任公司 | Parking charge method and device |
| CN107392313A (en) * | 2017-06-12 | 2017-11-24 | 五邑大学 | A kind of rail recognition methods based on deep learning |
| CN107392313B (en) * | 2017-06-12 | 2020-09-29 | 五邑大学 | Steel rail identification method based on deep learning |
| CN108052866A (en) * | 2017-11-17 | 2018-05-18 | 克立司帝控制系统（上海）有限公司 | Car license recognition learning method and system based on artificial neural network |
| CN108320531A (en) * | 2018-04-04 | 2018-07-24 | 武汉市技领科技有限公司 | A kind of speed measuring equipment and velocity-measuring system |
| CN109615879A (en) * | 2018-12-28 | 2019-04-12 | 成都路行通信息技术有限公司 | Speed abnormity early warning model, method and model building method based on car networking |
| CN109615879B (en) * | 2018-12-28 | 2020-12-11 | 成都路行通信息技术有限公司 | Vehicle speed abnormity early warning model and method based on Internet of vehicles and model construction method |
| CN109767629A (en) * | 2019-02-01 | 2019-05-17 | 深圳市润泽材料科技有限公司 | A kind of camera system and its speed-measuring method of integrated speed measuring function |
| CN110632339A (en) * | 2019-10-09 | 2019-12-31 | 天津天地伟业信息系统集成有限公司 | Water flow testing method of video flow velocity tester |
| CN110705493A (en) * | 2019-10-10 | 2020-01-17 | 深圳市元征科技股份有限公司 | Method and system for detecting vehicle running environment, electronic device and storage medium |
| CN112199545A (en) * | 2020-11-23 | 2021-01-08 | 湖南蚁坊软件股份有限公司 | Keyword display method and device based on picture character positioning and storage medium |
| CN113030506A (en) * | 2021-03-25 | 2021-06-25 | 上海其高电子科技有限公司 | Micro-area speed measurement method and system based on multi-license plate calibration library |
| CN113030506B (en) * | 2021-03-25 | 2022-07-12 | 上海其高电子科技有限公司 | Micro-area speed measurement method and system based on multi-license plate calibration library |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100452110C (en) | 2009-01-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101105893A (en) | Automobile video frequency discrimination speed-testing method | |
| CN111368687B (en) | Sidewalk vehicle illegal parking detection method based on target detection and semantic segmentation | |
| US8509486B2 (en) | Vehicle license plate recognition method and system thereof | |
| CN106652465B (en) | Method and system for identifying abnormal driving behaviors on road | |
| Alefs et al. | Road sign detection from edge orientation histograms | |
| CN109299674B (en) | Tunnel illegal lane change detection method based on car lamp | |
| CN105930791A (en) | Road traffic sign identification method with multiple-camera integration based on DS evidence theory | |
| CN105005771A (en) | Method for detecting full line of lane based on optical flow point locus statistics | |
| CN111382704A (en) | Vehicle line-pressing violation judgment method and device based on deep learning and storage medium | |
| CN103295420A (en) | Method for recognizing lane line | |
| Cho et al. | License plate extraction method for identification of vehicle violations at a railway level crossing | |
| CN113553916B (en) | Orbit dangerous area obstacle detection method based on convolutional neural network | |
| CN111783700B (en) | Automatic recognition and early warning method and system for pavement foreign matters | |
| CN109409191A (en) | A kind of zebra stripes vehicle evacuation detection method and system based on machine learning | |
| CN113903008A (en) | Ramp exit vehicle violation identification method based on deep learning and trajectory tracking | |
| CN106530825A (en) | Method for detecting traffic conflict between motor-assisted bicycle and automobile based on ST-MRF model | |
| CN106951820B (en) | Passenger flow statistical method based on annular template and ellipse fitting | |
| CN113657305B (en) | Video-based intelligent detection method for black smoke vehicle and ringeman blackness level | |
| CN107862341A (en) | A kind of vehicle checking method | |
| CN117294818A (en) | Building site panoramic monitoring method for airport construction | |
| CN109977772B (en) | Zebra crossing concession behavior detection method based on vehicle-mounted automobile data recorder video | |
| CN110909745A (en) | Train disinfection channel identification system | |
| Yao et al. | Fuzzy c-means image segmentation approach for axle-based vehicle classification | |
| CN114299438A (en) | Tunnel parking event detection method integrating traditional parking detection and neural network | |
| CN112949595A (en) | Improved pedestrian and vehicle safety distance detection algorithm based on YOLOv5 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20090114 Termination date: 20120714 |