CN102496019A - License plate character segmenting method - Google Patents
License plate character segmenting method Download PDFInfo
- Publication number
- CN102496019A CN102496019A CN2011104052270A CN201110405227A CN102496019A CN 102496019 A CN102496019 A CN 102496019A CN 2011104052270 A CN2011104052270 A CN 2011104052270A CN 201110405227 A CN201110405227 A CN 201110405227A CN 102496019 A CN102496019 A CN 102496019A
- Authority
- CN
- China
- Prior art keywords
- license plate
- segmentation
- character
- height
- region
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 69
- 230000011218 segmentation Effects 0.000 claims abstract description 96
- 238000003708 edge detection Methods 0.000 claims description 12
- 230000009466 transformation Effects 0.000 claims description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 6
- 238000005192 partition Methods 0.000 claims description 4
- 230000002452 interceptive effect Effects 0.000 claims description 2
- 238000000926 separation method Methods 0.000 claims description 2
- 238000001914 filtration Methods 0.000 claims 1
- 238000004364 calculation method Methods 0.000 abstract description 7
- 238000000638 solvent extraction Methods 0.000 abstract 1
- 238000001514 detection method Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 6
- 238000012937 correction Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 239000011800 void material Substances 0.000 description 4
- 238000007621 cluster analysis Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 230000009191 jumping Effects 0.000 description 2
- 238000003909 pattern recognition Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000005299 abrasion Methods 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000002902 bimodal effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
Images
Landscapes
- Character Input (AREA)
Abstract
The invention discloses a license plate character segmenting method which comprises the following steps of: S10. correcting an inclined license plate; S11. removing an upper frame and a lower frame of the license plate; S12. segmenting license plate characters; and S13. normalizing the character size. The license plate character segmenting method overcomes the shortcomings that the traditional license plate character segmenting method is greatly affected by the background of the license plate, is high in the partitioning error rate, huge in calculation amount and long in execution time; the license plate character segmenting method is based on vertical projections and center-to-center spacing of the characters; and according to the license plate character segmenting method, firstly, a vertical projection method is used for realizing the coarse segmentation of the license plate characters, and then a first or second character segmentation region is found according to the center-to-center spacing of the license plate characters, the widths of the license plate characters and the prior knowledge of the license plate, and finally the merging of Chinese characters and the fine segmentation of numbers and English characters are realized. The license plate character segmenting method has the advantages that the computation complexity is low, the capacity of resisting disturbance is strong, the characters can be segmented from the background and the reliability is very high.
Description
Technical Field
The invention belongs to the technical field of license plate recognition in the aspect of intelligent transportation, and particularly relates to a license plate character segmentation method.
Background
With the rapid development of road construction and the increasing number of automobiles in various countries, the task of traffic management is increasingly heavy, and the automatic detection and identification of automobile license plate numbers by using a computer automobile license plate identification technology plays a very important role in modern traffic monitoring. The license plate detection and recognition technology is one of important research subjects of a digital image processing and pattern recognition technology in the field of Intelligent Transportation Systems (ITS), plays an important role in promoting the development of Intelligent Transportation systems and the development of the Transportation industry, and has a wide market prospect.
As shown in fig. 1, the license plate detection and recognition technology is divided into four steps: license plate positioning, license plate correction, character segmentation and character recognition. The license plate character segmentation is an important component in the license plate detection and recognition technology, and the target character features can be further extracted and recognized only after the segmentation is effectively completed.
The currently common license plate character segmentation methods include a projection method, a connected domain method, a cluster analysis method and the like. The projection method is the most common license plate character segmentation method at present, and the idea is to perform projection in the vertical direction on a license plate image after binarization according to the characteristics of characters. Because the pixel points of the characters are more and concentrated, and each license plate character is separated by a certain gap. The projected image obtained by projection has seven relatively concentrated projection peak value groups, and then the characters of the license plate can be obtained by segmenting according to the lowest points among the peak values. Since the projection of the character block in the vertical direction not only obtains the local minimum value between characters, but also obtains the local minimum value in the gap in the character, the traditional projection segmentation method can easily segment the Chinese character into two parts or three parts, resulting in segmentation errors. In addition, the space points between the left and right frames of the license plate and the two or three characters can interfere with the projection segmentation, so that segmentation errors are caused; and the segmentation effect on characters in the images shot under different illumination conditions is poor, and the anti-jamming capability is poor.
The connected domain method is to extract all connected domains containing the initial points in the image by taking the target pixels on the horizontal line of the image as the initial points and growing the regions. Because the letters and the numbers in the license plate are written by one stroke, namely only one connected branch is included, each connected domain is a character, and the rest parts (the parts which are not accessed in the area growing process) in the image are removed as noise. The method has high requirements for removing noise interference, because the phenomenon of adhesion of characters and the edge of a license plate is very common (particularly through rivets at the second and sixth characters), which leads to extraction of a plurality of characters as one character and results in segmentation errors. In addition, many Chinese characters comprise a plurality of connected domains after binarization, and the stroke fracture phenomenon can occur after the binarization of numbers and letters, so partial character information can be lost by the method, and errors can easily occur in character segmentation.
The cluster analysis method is to realize the segmentation of the license plate characters by using a cluster analysis algorithm in pattern recognition. The method can better solve the problem that the Chinese characters are not communicated, and better solve the problems of noise interference existing in the character segmentation of the license plate, character adhesion caused by the abrasion of the license plate and the like. In addition, the coordinates of the preset class center are changed, so that some novel license plates can be better processed. However, the method has the disadvantages of complex program logic design, more loop nesting and overlong real-time processing time. Meanwhile, in order to improve the precision of the preset center, the width of the license plate is limited to a certain extent.
It can be seen that the license plate character segmentation method in the prior art has many defects: the traditional projection method and the connected domain method are greatly influenced by the background of the license plate, and when the characters of the license plate are polluted, the segmentation error rate is high, so that the identification of the characters of the license plate at the back is directly influenced; although the clustering analysis method can overcome a few slightly complex backgrounds, the calculation amount is large, and the requirement on the speed of a computer is high; the execution time is long, the time is consumed, and the margin is easily segmented into characters for the license plate image with the wide margin, so that misjudgment is caused.
Therefore, research is needed, and a technical scheme which has good real-time performance and low calculation complexity and can accurately segment characters from the background of the complex license plate is provided.
Disclosure of Invention
In order to solve the above problems, an object of the present invention is to provide a license plate character segmentation method with good real-time performance and low computation complexity, which can accurately segment characters from a complex license plate background.
In order to achieve the purpose, the technical scheme of the invention is as follows:
a license plate character segmentation method comprises the following steps:
s10: correcting the inclination of the license plate;
s11: removing the upper and lower frames of the license plate;
s12: segmenting license plate characters;
s13: and (5) normalizing the character size.
Further, the step S10 includes the following steps:
s100: canny edge detection is carried out on the original gray license plate image obtained by positioning;
s101: detecting a straight line by using probability Hough transformation on the license plate image subjected to Canny edge detection, and calculating the inclination angle of the license plate;
s102: and rotating the original gray-scale license plate image by a corresponding angle according to the license plate inclination angle obtained in the step S101 to obtain a horizontal gray-scale license plate image.
Further, the step S11 includes the following steps:
s110: binarizing the horizontal gray level license plate image obtained in the step S102 by using an Otsu threshold value method to obtain a binary license plate image;
s111: and removing the upper and lower frames of the license plate of the binary license plate image by using a gray level jump method.
Further, the step S12 includes the following steps:
s120: carrying out vertical projection on the binary license plate image without the upper and lower frames of the license plate, and then carrying out rough segmentation;
s121: and finely dividing the license plate image after the coarse division by using the character center distance and the license plate priori knowledge.
The license plate character segmentation method overcomes the defects that the existing license plate character segmentation method is greatly influenced by the background of the license plate, has high segmentation error rate, large calculation amount and long execution time, and is based on the license plate character segmentation of vertical projection and character center spacing. The license plate character segmentation method is low in calculation complexity, high in anti-interference capacity, capable of segmenting characters from a complex background and high in reliability.
Drawings
Fig. 1 is a flow chart of license plate detection and recognition in the prior art.
FIG. 2 is a flow chart of the character segmentation process of the present invention.
FIG. 3 is an original gray-scale license plate image of the present invention using 7-bit characters as an example.
Fig. 4 is a license plate image after Canny edge detection in fig. 3.
Fig. 5 is a straight line detected by the probability Hough transform of fig. 3.
Fig. 6 is a license plate image after the inclination correction of fig. 3.
FIG. 7 is the license plate image after the binarization by the Otsu method in FIG. 3.
Fig. 8 is a license plate image of fig. 3 with upper and lower rims removed.
FIG. 9 is a license plate image of the character of FIG. 3 after segmentation.
FIG. 10 is a graphical representation of the license plate characters of FIG. 3 after being segmented and normalized.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 2, the license plate character segmentation method of the present invention includes the following steps:
s10: correcting the inclination of the license plate;
s11: removing the upper and lower frames of the license plate;
s12: segmenting license plate characters;
s13: and (5) normalizing the character size.
Wherein,
the step S10 of license plate inclination correction comprises the following steps:
s100: canny edge detection is carried out on the original gray license plate image obtained by positioning;
s101: detecting a straight line by using probability Hough transformation on the license plate image subjected to Canny edge detection, and calculating the inclination angle of the license plate;
s102: and rotating the original gray-scale license plate image by a corresponding angle according to the license plate inclination angle obtained in the step S101 to obtain a horizontal gray-scale license plate image.
Step S11 license plate upper and lower frame removing includes the following steps:
s110: binarizing the horizontal gray level license plate image obtained in the step S102 by using an Otsu threshold value method to obtain a binary license plate image;
s111: and removing the upper and lower frames of the license plate of the binary license plate image by using a gray level jump method.
The step S12 of segmenting the license plate characters comprises the following steps:
s120: carrying out vertical projection on the binary license plate image without the upper and lower frames of the license plate, and then carrying out rough segmentation;
s121: and finely dividing the license plate image after the coarse division by using the character center distance, the license plate character width and the license plate priori knowledge. The license plate priori knowledge mainly comprises the aspect ratio of characters of an actual license plate and the characters of the first and second license plates, which do not comprise the character 1.
Referring to fig. 3 to 10, an embodiment of the present invention takes a license plate jing LR8251 as an example for description.
S10: license plate tilt correction
In an ideal situation, the license plate image should be a rectangle, but due to the influence of the angle between the lens and the license plate, the movement of the vehicle, the road surface condition, etc. during shooting, the license plate often inclines to a certain degree and may also be deformed. The tilted license plate will then affect the vertical projection of the license plate, thereby indirectly causing character segmentation errors. Therefore, it is necessary to correct the tilted license plate image. The license plate inclination corrector comprises the following specific components:
s100: and carrying out Canny edge detection on the positioned original gray license plate image. The inclination angle of the license plate is calculated according to the inclination of the upper frame and the lower frame of the license plate, so that the Canny edge detection method is used for highlighting the edge of the original gray-scale license plate image so as to be convenient for detecting the frame straight line in the next step. In this embodiment, the open source computer vision library OpenCV of Intel corporation is used, and the function prototype of Canny edge detection is as follows:
void cvCanny(const CvArr*image，CvArr*edges，double threshold1，doublethreshold2，int aperture_size＝3)；
where image represents the input image, edges represents the output image, a small threshold of threshold1 and threshold is used to control edge linking, a large threshold is used to control the initial segmentation of the edge, and aperture _ size is the Sobel operator kernel size. The functions and parameters used in this example are as follows:
cvCanny(image，edges，50，200，3)；
the effect of the license plate image after Canny edge detection is shown in fig. 4.
S101: and (3) detecting a straight line by using probability Hough transformation on the license plate image subjected to Canny edge detection, and calculating the inclination angle of the license plate. The line detection usually uses the traditional Hough transformation algorithm, but the algorithm needs to scan and calculate each pixel of the whole image, so that the calculation amount is large, and the processing time is long; in the invention, probability Hough transformation is adopted to detect straight lines in the image so as to reduce the calculation time. The functional prototype of the probabilistic Hough transform in OpenCV is as follows:
CvSeq*cvHoughLines2(CvArr*image，void*line_storage，int method，doublerho，double theta，int threshold，double param1＝0，double param2＝0)；
the image is an input image, the line _ storage is a storage bin of a detected line segment, the method is a Hough transformation variable, rho is the distance precision of a unit related to a pixel, theta is the angle precision of radian measurement, threshold is a threshold parameter, param1 is the minimum line segment length, and param2 represents the maximum interval for connecting broken line segments on the same straight line. The functions and parameters used in this example are as follows:
cvHoughLines2(image，lines_storage，CV_HOUGH_PROBABILISTIC，1，CV_PI/180，80，30，10)；
after the line segments are detected by adopting the function, the slope and the inclination angle of each line segment are calculated, and the largest angle is selected from the inclination angles smaller than 30 degrees as the inclination angle of the license plate. The line segments detected by the probabilistic Hough transform are shown in fig. 5.
S102: and rotating the original gray license plate image by a corresponding angle according to the detected inclination angle to obtain a horizontal gray license plate image. In the embodiment, the original gray-scale license plate image is rotated by adopting a bilinear interpolation algorithm, so that the details in the rotated image can be reserved, the effect of the rotated image is as shown in fig. 6, and the license plate characters are not deformed after the rotation.
S11: removing upper and lower frames of license plate
The license plate still can have the interference of upper and lower frame after slope correction, if do not remove, will bring adverse effect to character cutting and discernment. The removal of the upper and lower frames of the license plate is as follows:
s110: and carrying out binarization on the obtained horizontal gray level license plate image by using an Otsu threshold value method to obtain a binary license plate image. In this embodiment, the image is divided into a background part and a foreground part according to the gray characteristic of the image. The Otsu threshold method obtains a segmentation threshold by calculating the maximum inter-class variance, and can obtain a better segmentation effect by using the Otsu threshold method under an ideal 'bimodal condition'. The binarization function prototype in OpenCV is as follows:
void cvThreshold(const CvArr*src，CvArr*dst，double threshold，doublemax_value，int threshold_type)；
where src is the original array, dst is the output array, threshold is the threshold, max _ value is the maximum value using CV _ THRESH _ BINARY and CV _ THRESH _ BINARY _ INV, and threshold _ type is the threshold type. The functions and parameters used in this example are as follows:
cvThreshold(src，dst，100，255，CV_THRESH_OTSU)；
as shown in fig. 7, the binary image is obtained by binarizing the corrected gray-scale license plate image by the Otsu threshold method.
S111: and removing the upper and lower frames of the license plate of the binary license plate image by using a gray level jump method. According to the prior knowledge of the license plate, 7 characters exist on the license plate after binarization, the number of the jumping points of each character is at least 2, the total number of the horizontal black and white jumping points in a character area is more than 14, and the non-character area does not meet the characteristic. Therefore, the method for removing the upper frame and the lower frame of the license plate by using the black-and-white jump method comprises the following steps:
s1110: scanning upwards from one half of the binary image, counting the black and white jump number of each line, if the value of the black and white jump number is less than 14, indicating that the line is not a character area, and taking the line as the first line of the character area;
s1111: scanning downwards from one half of the binary image, counting the black and white jump number of each line, if the value of the black and white jump number is less than 14, indicating that the line is not a character area, and taking the line as the last line of the character area;
s1112: the area between the first line and the last line is the character area, thus eliminating the influence of the upper and lower frames.
As shown in fig. 8, it shows an effect diagram of the license plate image after the frame removal operation.
S12: license plate character segmentation
S120: and carrying out vertical projection on the binary license plate image without the upper and lower frames of the license plate, and then carrying out rough segmentation. The method comprises the following specific steps:
s1200: starting column scanning from the left side of the binary license plate image after removing the upper frame and the lower frame of the license plate, recording a column number a when the number of black pixels is less than a certain threshold value T (2 in the embodiment), and putting the column number a into an array ColNo [2i ];
s1201: if the number of black pixels is larger than a certain threshold (2 in the embodiment), recording the column number b, and putting the column number b into an array ColNo [2i +1 ]; combining the a obtained in step S1200, a first divided region can be determined, where a and b are the left and right boundaries of the divided region, respectively;
s1202: the scanning is continued according to the above steps S1200 and S1201 until all columns are scanned, and then the even positions in the array ColNo [ i ] store the left boundary of the divided area, and the odd positions store the right boundary of the divided area.
S121: and finely dividing the license plate image after the coarse division by using the character center distance and the license plate priori knowledge. Taking the actual license plate height of a domestic common car as an example, the actual license plate height of the domestic common car is 140mm, and the actual license plate width of the domestic common car is 440 mm; the actual character height is 90mm, and the actual width of other individual characters except for "1" is 45 mm; the actual character center-to-center spacing was 57mm (two-to-three character center-to-center spacing was 79 mm). Therefore, the ratio of the width to the height of the actual characters is 1: 2, and the ratio of the center-to-center distance of the actual characters to the height of the license plate characters is 57: 90 and 79: 90 (the ratio of the center-to-center distance of the two or three characters to the height of the license plate characters). Therefore, under the condition of knowing the height of the license plate characters, the width range of the characters and the range of the character center distance can be determined according to the width-to-height ratio of the actual license plate and the ratio of the character center distance to the height of the license plate characters. The method can realize the fine segmentation of the license plate characters by utilizing the character center distance and the license plate priori knowledge, and specifically comprises the following steps:
s1210: the number of roughly divided regions is checked. If the number is less than 7, proceeding to the subsequent step S1211; if the number is more than 15, the segmentation fails, and the program exits; and proceeds to subsequent step S1212 if intermediate between 7 and 15.
S1211: after increasing the threshold, the coarse segmentation is performed again and the number of regions is checked. If the number is less than 7 or greater than 15, the segmentation fails and the procedure exits.
S1212: the width of each segmented area is checked and compared with a set threshold MaxThresholdWidth (the threshold can be determined according to the positioned height (height), which is taken as height/2+ height/10+2.5 in the embodiment). If the value is larger than the threshold value, the threshold value is increased again, and then the image is roughly segmented again; if the rough segmentation is not performed, the flow proceeds directly to the subsequent step S1214.
S1213: the number of segmented regions is checked. If the number is larger than 15, the segmentation fails, and the program exits.
S1214: the width of each divided region is checked. If the width of the region is larger than the MaxThresholdWidth, the segmentation fails and the program exits.
S1215: and counting the number of white pixel points in each partition area. If the number of the divided areas is less than 7, the division fails, and the program exits; if the number of white pixels in the partition area is less than a threshold (height 2 in this embodiment), the area is discarded. So that small disturbing points and the separation between the second and third characters can be filtered out.
S1216: the width of each segmented region is checked. If its width is less than a certain threshold (here set to height/10-0.5), it is removed. This operation may remove some of the interfering segmented regions but ensures that the character "1" is not removed.
S1217: checking the number of the divided areas: if the number of the segmentation modules is less than 7, the segmentation is failed, and the program is exited.
S1218: and searching a next segmentation region omega with the region width meeting a certain range (height/2-height/10-1.5, height/2+ height/10+1.5), and if the search is not completed, failing to segment, and exiting the program.
S1219: the horizontal distance between the center position of the segment Ω and the center position of the next segment is checked to see if it is equal to T1 (license plate height 57/90) or T2 (license plate height 79/90). If the number is equal to T1, taking the segmentation region omega as a first license plate character region, and proceeding to step S1220; if the number is equal to T2, taking the segmentation area omega as a second character area of the license plate, and entering step S1221; if there is no match with both T1 and T2, the next divided region Ω whose region width satisfies a certain range is searched for, and step S1219 is repeatedly performed.
S1220: the found segmentation region omega is the first license plate character region (Chinese character region), the number of the segmentation regions behind the first license plate character region is judged, if the number is more than or equal to 6, the six following regions are sequentially read to form the license plate character region, the segmentation is finished, and the program exits; if the number is less than 6, the segmentation fails and the procedure exits.
S1221: and if the number of the found divided areas is less than 5, the division fails, and the program exits.
S1222: the number of segmented regions in front of the examination region Ω. If the number is less than 2, the segmentation fails, and the program exits; if equal to 2, go to step S1223; if 3 or more, the flow proceeds to step S1224.
S1223: the two regions before region omega are merged and the merged region width is checked. If the width of the merged region meets a certain range (height/2-height/10-1.5, height/2+ height/10+1.5 in the embodiment), and the horizontal distance between the merged region and the center of the region omega is approximately the height of the license plate 57/90, the merged region is considered as the Chinese character region. Sequentially reading five regions behind the region omega to form a license plate character region, finishing segmentation, and exiting the program; if one of the above requirements is not met, the segmentation fails and the procedure exits.
S1224: the two regions in front of region omega are merged first and the merged region width is checked. If the width of the merged region meets a certain range (height/2-height/10-1.5, height/2+ height/10+1.5 in the embodiment), and the horizontal distance between the merged region and the center of the region omega is approximately the height of the license plate 57/90, the merged region is considered as the Chinese character region. Sequentially reading five regions behind the region omega to form a license plate character region, finishing segmentation, and exiting the program; if the combined width is larger than (height/2+ height/10+1.5), the segmentation fails, and the program exits; if the merged width is less than (height/2-height/10-1.5), the process proceeds to step S1225.
S1225: the three regions in front of region Ω are merged and the merged region width is checked. If the width of the merged region meets a certain range (height/2-height/10-1.5, height/2+ height/10+1.5 in the embodiment), and the horizontal distance between the merged region and the center of the region omega is approximately the height of the license plate 57/90, the merged region is considered as the Chinese character region. Sequentially reading five regions behind the region omega to form a license plate character region, finishing segmentation, and exiting the program; if one of the above requirements is not met, the segmentation fails and the procedure exits.
As shown in fig. 9, it shows a license plate image after character segmentation.
S13: character size normalization
The size of the license plate image often causes that the size of the segmented characters is also different, and in order to facilitate subsequent character recognition, the size of the segmented characters needs to be normalized. The characters are normalized to a size of 36x20 in this embodiment. In OpenCV, the function prototype that changes the image size is as follows:
Void cvResize(const CvArr*src，CvArr*dst，int interpolation＝CV_INTER_LINEAR)；
src is the input image, dst is the output image, and interpolation is the interpolation method. The functions and parameters used in this example are as follows:
cvResize(src，dst，CV_INTER_NN)。
as shown in fig. 10, a graphical representation of the license plate characters after being segmented and normalized is shown.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (9)
1. A license plate character segmentation method is characterized by comprising the following steps:
s10: correcting the inclination of the license plate;
s11: removing the upper and lower frames of the license plate;
s12: segmenting license plate characters;
s13: and (5) normalizing the character size.
2. The license plate character segmentation method of claim 1, wherein: the step S10 includes the following steps:
s100: canny edge detection is carried out on the original gray license plate image obtained by positioning;
s101: detecting a straight line by using probability Hough transformation on the license plate image subjected to Canny edge detection, and calculating the inclination angle of the license plate;
s102: and rotating the original gray-scale license plate image by a corresponding angle according to the license plate inclination angle obtained in the step S101 to obtain a horizontal gray-scale license plate image.
3. The license plate character segmentation method of claim 2, wherein: the step S11 includes the following steps:
s110: binarizing the horizontal gray level license plate image obtained in the step S102 by using an Otsu threshold value method to obtain a binary license plate image;
s111: and removing the upper and lower frames of the license plate of the binary license plate image by using a gray level jump method.
4. The license plate character segmentation method of claim 3, wherein: the step S12 includes the following steps:
s120: carrying out vertical projection on the binary license plate image without the upper and lower frames of the license plate, and then carrying out rough segmentation;
s121: and finely dividing the license plate image after the coarse division by using the character center distance and the license plate priori knowledge.
5. The license plate character segmentation method of claim 4, wherein: in step S102, the original gray-scale license plate image is rotated by using a bilinear interpolation algorithm.
6. The license plate character segmentation method of claim 5, wherein: in step S110, the Otsu algorithm is used to divide the image into two parts, namely, a background part and a foreground part according to the gray characteristics of the image.
7. The license plate character segmentation method of claim 6, wherein: the step S111 includes the steps of:
s1110: scanning upwards from one half of the binary image, counting the black and white jump number of each line, if the value of the black and white jump number is less than 14, indicating that the line is not a character area, and taking the line as the first line of the character area;
s1111: scanning downwards from one half of the binary image, counting the black and white jump number of each line, if the value of the black and white jump number is less than 14, indicating that the line is not a character area, and taking the line as the last line of the character area;
s1112: the area between the first line and the last line is the character area, thus eliminating the influence of the upper and lower frames.
8. The license plate character segmentation method of claim 7, wherein: the step S120 includes the steps of:
s1200: starting column scanning from the left side of the binary license plate image after removing the upper and lower frames of the license plate, recording the column number a when the number of black pixels is less than a certain threshold value T, and putting the column number a into an array ColNo [2i ];
s1201: if the number of black pixels is larger than a certain threshold value, recording the column number b, and putting the column number b into an array ColNo [2i +1 ]; combining the a obtained in step S1200, a first divided region can be determined, where a and b are the left and right boundaries of the divided region, respectively;
s1202: the scanning is continued according to the above steps S1200 and S1201 until all columns are scanned, and then the even positions in the array ColNo [ i ] store the left boundary of the divided area, and the odd positions store the right boundary of the divided area.
9. The license plate character segmentation method of claim 8, wherein: the step S121 includes the steps of:
s1210: checking the number of the roughly divided regions, and if the number is less than 7, proceeding to the subsequent step S1211; if the number is more than 15, the segmentation fails, and the program exits; and if intermediate between 7 and 15, proceeds to subsequent step S1212;
s1211: after the threshold value is increased, performing rough segmentation again, checking the number of the regions, if the number is less than 7 or more than 15, failing in segmentation, and exiting the program;
s1212: checking the width of each segmentation area, comparing the width with a set threshold MaxThresholdWidth, if the width is larger than the threshold, increasing the threshold again, and then performing coarse segmentation on the image again; if the rough segmentation is not performed, the flow proceeds directly to the subsequent step S1214.
S1213: checking the number of the divided areas, if the number is more than 15, failing to divide, and exiting the program;
s1214: checking the width of each segmentation region, if the width of the region is larger than MaxThresholdWidth, failing to segment, and exiting the program;
s1215: counting the number of white pixel points in each partition region, and if the number of the white pixel points in each partition region is less than a threshold, abandoning the region, thereby filtering small interference points and the separation point between the second character and the third character;
s1216: checking the width of each segmentation region and removing it if its width is less than a certain threshold, which removes some of the interfering segmentation regions but ensures that the character "1" is not removed;
s1217: checking the number of the divided areas: if the number of the segmentation is less than 7, the segmentation fails, and the program exits;
s1218: searching a next segmentation region omega with the width between height/2-height/10-1.5 and height/2+ height/10+1.5, and if the next segmentation region omega cannot be searched, failing to segment, and exiting the program;
s1219: checking the horizontal distance between the center position of the divided region omega and the center position of the next divided region, and checking whether the distance is equal to T1 (license plate height) 57/90 or T2 (license plate height) 79/90; if the number is equal to T1, taking the segmentation region omega as a first license plate character region, and proceeding to step S1220; if the number is equal to T2, taking the segmentation area omega as a second character area of the license plate, and entering step S1221; if the current region width does not meet both the T1 and the T2, searching a next divided region Ω whose region width satisfies a certain range, and repeatedly executing step S1219;
s1220: if the number of the found division regions omega is greater than or equal to 6, reading the next six regions in sequence to form the license plate character region, finishing the division, and exiting the program; if the number is less than 6, the segmentation fails, and the program exits;
s1221: if the number of the found divided areas is less than 5, the division fails, and the program exits;
s1222: checking the number of the segmentation areas in front of the area omega, if the number is less than 2, failing to segment, and exiting the program; if equal to 2, go to step S1223; if not, go to step S1224;
s1223: merging two areas in front of the area omega, checking the width of the merged area, if the width is between height/2-height/10-1.5 and height/2+ height/10+1.5, and the horizontal distance between the merged area and the center of the area omega is approximately the height of the license plate 57/90, considering the merged area as a Chinese character area, sequentially reading five areas behind the area omega to form a license plate character area, finishing segmentation, and exiting the program; if one of the requirements is not met, the segmentation fails, and the program exits;
s1224: merging two areas in front of an area omega, checking the width of the merged area, if the width of the merged area is between height/2-height/10-1.5 and height/2+ height/10+1.5, and the horizontal distance between the merged area and the center of the area omega is approximately the height of the license plate 57/90, considering the merged area as a Chinese character area, sequentially reading five areas behind the area omega to form a license plate character area, finishing segmentation, and exiting the program; if the combined width is greater than height/2+ height/10+1.5, the segmentation fails, and the program exits; if the combined width is less than height/2-height/10-1.5, go to step S1225;
s1225: merging the three areas in front of the area omega, and checking the width of the merged area; if the width of the combined region is between height/2-height/10-1.5 and height/2+ height/10+1.5, and the horizontal distance between the combined region and the center of the region omega is approximately the height of the license plate 57/90, the combined region is considered as a Chinese character region; sequentially reading five regions behind the region omega to form a license plate character region, finishing segmentation, and exiting the program; if one of the above requirements is not met, the segmentation fails and the procedure exits.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011104052270A CN102496019A (en) | 2011-12-08 | 2011-12-08 | License plate character segmenting method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011104052270A CN102496019A (en) | 2011-12-08 | 2011-12-08 | License plate character segmenting method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102496019A true CN102496019A (en) | 2012-06-13 |
Family
ID=46187844
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2011104052270A Pending CN102496019A (en) | 2011-12-08 | 2011-12-08 | License plate character segmenting method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102496019A (en) |
Cited By (25)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103077388A (en) * | 2012-10-31 | 2013-05-01 | 浙江大学 | Rapid text scanning method oriented to portable computing equipment |
| CN103198315A (en) * | 2013-04-17 | 2013-07-10 | 南京理工大学 | License plate character segmentation algorithm based on character outline and template matching |
| CN103455815A (en) * | 2013-08-27 | 2013-12-18 | 电子科技大学 | Self-adaptive license plate character segmentation method in complex scene |
| CN104657726A (en) * | 2015-03-18 | 2015-05-27 | 哈尔滨工程大学 | Identification method for truck numbers of railway tank trucks |
| CN104700432A (en) * | 2015-03-24 | 2015-06-10 | 银江股份有限公司 | Self-adaptive adhered vehicle separating method |
| CN105117727A (en) * | 2015-09-18 | 2015-12-02 | 浙江工业大学 | Method for fast positioning license plate |
| CN105608455A (en) * | 2015-12-18 | 2016-05-25 | 浙江宇视科技有限公司 | License plate tilt correction method and apparatus |
| CN105631425A (en) * | 2015-12-29 | 2016-06-01 | 厦门科拓通讯技术股份有限公司 | License plate recognition method based on video streams and system thereof and intelligent digital video camera |
| CN106529540A (en) * | 2016-11-07 | 2017-03-22 | 湖南源信光电科技有限公司 | Gray projection-based license plate character segmentation method |
| CN106529534A (en) * | 2016-11-07 | 2017-03-22 | 湖南源信光电科技有限公司 | Variable-length license plate character segmentation method based on hybrid tilt correction and projection method |
| CN106951896A (en) * | 2017-02-22 | 2017-07-14 | 武汉黄丫智能科技发展有限公司 | A kind of license plate image sloped correcting method |
| CN106980857A (en) * | 2017-02-24 | 2017-07-25 | 浙江工业大学 | A kind of Brush calligraphy segmentation recognition method based on rubbings |
| CN107992785A (en) * | 2016-10-26 | 2018-05-04 | 北京君正集成电路股份有限公司 | The recognition methods of fuzzy license plate and device |
| CN104834930B (en) * | 2015-05-25 | 2018-07-24 | 中孚信息股份有限公司 | A kind of image method for looking up Chinese characters for multisystem platform |
| CN108446702A (en) * | 2018-03-14 | 2018-08-24 | 深圳怡化电脑股份有限公司 | A kind of image character dividing method, device, equipment and storage medium |
| CN108470175A (en) * | 2018-03-27 | 2018-08-31 | 康体佳智能科技（深圳）有限公司 | Vehicle License Plate Recognition System based on neural network and recognition methods |
| CN109034145A (en) * | 2018-07-06 | 2018-12-18 | 西安电子科技大学 | Bank's card number recognition methods based on OpenCV |
| CN109492562A (en) * | 2018-10-30 | 2019-03-19 | 河南科技大学 | A kind of contactless domestic animal identifying system |
| CN110059695A (en) * | 2019-04-23 | 2019-07-26 | 厦门商集网络科技有限责任公司 | A kind of character segmentation method and terminal based on upright projection |
| CN110335286A (en) * | 2019-07-11 | 2019-10-15 | 京东方科技集团股份有限公司 | Acquisition methods, computer equipment and the computer-readable medium of gray threshold |
| CN111027546A (en) * | 2019-12-05 | 2020-04-17 | 北京嘉楠捷思信息技术有限公司 | Character segmentation method and device and computer readable storage medium |
| CN111223065A (en) * | 2020-01-13 | 2020-06-02 | 中国科学院重庆绿色智能技术研究院 | Image correction method, irregular text recognition method, device, storage medium and device |
| CN111598104A (en) * | 2020-06-30 | 2020-08-28 | 成都鹏业软件股份有限公司 | License plate character recognition method and system |
| CN112215852A (en) * | 2020-09-29 | 2021-01-12 | 忻州师范学院 | Digital image segmentation method based on cluster learning device integration |
| US11615634B2 (en) | 2021-07-01 | 2023-03-28 | International Business Machines Corporation | Character recognition of license plate under complex background |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020159636A1 (en) * | 2000-03-14 | 2002-10-31 | Lienhart Rainer W | Generalized text localization in images |
| CN101408933A (en) * | 2008-05-21 | 2009-04-15 | 浙江师范大学 | Method for recognizing license plate character based on wide gridding characteristic extraction and BP neural network |
-
2011
- 2011-12-08 CN CN2011104052270A patent/CN102496019A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020159636A1 (en) * | 2000-03-14 | 2002-10-31 | Lienhart Rainer W | Generalized text localization in images |
| CN101408933A (en) * | 2008-05-21 | 2009-04-15 | 浙江师范大学 | Method for recognizing license plate character based on wide gridding characteristic extraction and BP neural network |
Cited By (42)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103077388A (en) * | 2012-10-31 | 2013-05-01 | 浙江大学 | Rapid text scanning method oriented to portable computing equipment |
| CN103077388B (en) * | 2012-10-31 | 2016-01-20 | 浙江大学 | Fast text towards portable computing device sweeps the method for reading |
| CN103198315A (en) * | 2013-04-17 | 2013-07-10 | 南京理工大学 | License plate character segmentation algorithm based on character outline and template matching |
| CN103198315B (en) * | 2013-04-17 | 2015-12-02 | 南京理工大学 | Based on the Character Segmentation of License Plate of character outline and template matches |
| CN103455815B (en) * | 2013-08-27 | 2016-08-17 | 电子科技大学 | A kind of self adaptation registration number character dividing method under complex scene |
| CN103455815A (en) * | 2013-08-27 | 2013-12-18 | 电子科技大学 | Self-adaptive license plate character segmentation method in complex scene |
| CN104657726A (en) * | 2015-03-18 | 2015-05-27 | 哈尔滨工程大学 | Identification method for truck numbers of railway tank trucks |
| CN104700432A (en) * | 2015-03-24 | 2015-06-10 | 银江股份有限公司 | Self-adaptive adhered vehicle separating method |
| CN104700432B (en) * | 2015-03-24 | 2017-11-03 | 银江股份有限公司 | A kind of adaptive adhesion Method of Vehicle Segmentation |
| CN104834930B (en) * | 2015-05-25 | 2018-07-24 | 中孚信息股份有限公司 | A kind of image method for looking up Chinese characters for multisystem platform |
| CN105117727A (en) * | 2015-09-18 | 2015-12-02 | 浙江工业大学 | Method for fast positioning license plate |
| CN105117727B (en) * | 2015-09-18 | 2018-05-08 | 浙江工业大学 | A kind of Location Method of Vehicle License Plate |
| CN105608455B (en) * | 2015-12-18 | 2019-06-11 | 浙江宇视科技有限公司 | A kind of license plate tilt correction method and device |
| CN105608455A (en) * | 2015-12-18 | 2016-05-25 | 浙江宇视科技有限公司 | License plate tilt correction method and apparatus |
| CN105631425B (en) * | 2015-12-29 | 2020-04-07 | 厦门科拓通讯技术股份有限公司 | License plate recognition method and system based on video stream and intelligent digital camera |
| CN105631425A (en) * | 2015-12-29 | 2016-06-01 | 厦门科拓通讯技术股份有限公司 | License plate recognition method based on video streams and system thereof and intelligent digital video camera |
| CN107992785A (en) * | 2016-10-26 | 2018-05-04 | 北京君正集成电路股份有限公司 | The recognition methods of fuzzy license plate and device |
| CN107992785B (en) * | 2016-10-26 | 2021-08-31 | 北京君正集成电路股份有限公司 | Fuzzy license plate recognition method and device |
| CN106529540A (en) * | 2016-11-07 | 2017-03-22 | 湖南源信光电科技有限公司 | Gray projection-based license plate character segmentation method |
| CN106529534A (en) * | 2016-11-07 | 2017-03-22 | 湖南源信光电科技有限公司 | Variable-length license plate character segmentation method based on hybrid tilt correction and projection method |
| CN106951896B (en) * | 2017-02-22 | 2020-01-03 | 武汉黄丫智能科技发展有限公司 | License plate image tilt correction method |
| CN106951896A (en) * | 2017-02-22 | 2017-07-14 | 武汉黄丫智能科技发展有限公司 | A kind of license plate image sloped correcting method |
| CN106980857A (en) * | 2017-02-24 | 2017-07-25 | 浙江工业大学 | A kind of Brush calligraphy segmentation recognition method based on rubbings |
| CN106980857B (en) * | 2017-02-24 | 2020-05-05 | 浙江工业大学 | Chinese calligraphy segmentation and recognition method based on copybook |
| CN108446702B (en) * | 2018-03-14 | 2022-05-31 | 深圳怡化电脑股份有限公司 | Image character segmentation method, device, equipment and storage medium |
| CN108446702A (en) * | 2018-03-14 | 2018-08-24 | 深圳怡化电脑股份有限公司 | A kind of image character dividing method, device, equipment and storage medium |
| CN108470175A (en) * | 2018-03-27 | 2018-08-31 | 康体佳智能科技（深圳）有限公司 | Vehicle License Plate Recognition System based on neural network and recognition methods |
| CN109034145B (en) * | 2018-07-06 | 2021-11-09 | 西安电子科技大学 | Bank card number identification method based on OpenCV |
| CN109034145A (en) * | 2018-07-06 | 2018-12-18 | 西安电子科技大学 | Bank's card number recognition methods based on OpenCV |
| CN109492562B (en) * | 2018-10-30 | 2021-05-25 | 河南科技大学 | A contactless livestock identification system |
| CN109492562A (en) * | 2018-10-30 | 2019-03-19 | 河南科技大学 | A kind of contactless domestic animal identifying system |
| CN110059695B (en) * | 2019-04-23 | 2021-08-27 | 厦门商集网络科技有限责任公司 | Character segmentation method based on vertical projection and terminal |
| CN110059695A (en) * | 2019-04-23 | 2019-07-26 | 厦门商集网络科技有限责任公司 | A kind of character segmentation method and terminal based on upright projection |
| CN110335286A (en) * | 2019-07-11 | 2019-10-15 | 京东方科技集团股份有限公司 | Acquisition methods, computer equipment and the computer-readable medium of gray threshold |
| CN111027546A (en) * | 2019-12-05 | 2020-04-17 | 北京嘉楠捷思信息技术有限公司 | Character segmentation method and device and computer readable storage medium |
| CN111027546B (en) * | 2019-12-05 | 2024-03-26 | 嘉楠明芯(北京)科技有限公司 | Character segmentation method, device and computer readable storage medium |
| CN111223065A (en) * | 2020-01-13 | 2020-06-02 | 中国科学院重庆绿色智能技术研究院 | Image correction method, irregular text recognition method, device, storage medium and device |
| CN111223065B (en) * | 2020-01-13 | 2023-08-01 | 中国科学院重庆绿色智能技术研究院 | Image correction method, irregular text recognition method, device, storage medium and equipment |
| CN111598104A (en) * | 2020-06-30 | 2020-08-28 | 成都鹏业软件股份有限公司 | License plate character recognition method and system |
| CN111598104B (en) * | 2020-06-30 | 2023-05-12 | 成都鹏业软件股份有限公司 | License plate character recognition method and system |
| CN112215852A (en) * | 2020-09-29 | 2021-01-12 | 忻州师范学院 | Digital image segmentation method based on cluster learning device integration |
| US11615634B2 (en) | 2021-07-01 | 2023-03-28 | International Business Machines Corporation | Character recognition of license plate under complex background |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102496019A (en) | License plate character segmenting method | |
| CN109740469B (en) | Lane line detection method, lane line detection device, computer device, and storage medium | |
| Teoh et al. | Symmetry-based monocular vehicle detection system | |
| EP1085456B1 (en) | Character segmentation method for vehicle license plate recognition | |
| CN102722711B (en) | License plate character sequence identification method and device in license plate type identification | |
| CN102043959B (en) | License plate character segmentation method | |
| CN104809433B (en) | A kind of zebra line detecting method based on maximum stable region and stochastical sampling | |
| CN109886896A (en) | A blue license plate segmentation and correction method | |
| CN109583280A (en) | Lane detection method, apparatus, equipment and storage medium | |
| CN108537099A (en) | A kind of licence plate recognition method of complex background | |
| CN103699905A (en) | A license plate location method and device | |
| CN105303153A (en) | Vehicle license plate identification method and apparatus | |
| JP2019159606A (en) | Lane line recognition system | |
| US7986841B2 (en) | Method and apparatus for image edge detection | |
| EP3115933A1 (en) | Image processing device, image capturing device, mobile body control system, image processing method, and computer-readable recording medium | |
| CN114387591A (en) | License plate recognition method, system, device and storage medium | |
| CN111027544A (en) | MSER license plate positioning method and system based on visual saliency detection | |
| JP2015064752A (en) | Vehicle monitoring apparatus and vehicle monitoring method | |
| CN103745236A (en) | Texture image identification method and texture image identification device | |
| CN106778736A (en) | The licence plate recognition method and its system of a kind of robust | |
| CN107423735A (en) | It is a kind of to utilize horizontal gradient and the algorithm of locating license plate of vehicle of saturation degree | |
| CN111386530B (en) | Vehicle detection method and apparatus | |
| CN112101351B (en) | Text line rotation correction method and device based on projection | |
| CN105023002A (en) | Vehicle logo positioning method based on active vision | |
| CN113642553B (en) | Method for accurately positioning unconstrained license plate by combining whole and part target detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20120613 |