WO2013063820A1 - 一种车牌图像定位的方法和装置 - Google Patents

一种车牌图像定位的方法和装置 Download PDF

Info

Publication number
WO2013063820A1
WO2013063820A1 PCT/CN2011/081964 CN2011081964W WO2013063820A1 WO 2013063820 A1 WO2013063820 A1 WO 2013063820A1 CN 2011081964 W CN2011081964 W CN 2011081964W WO 2013063820 A1 WO2013063820 A1 WO 2013063820A1
Authority
WO
WIPO (PCT)
Prior art keywords
gray
license plate
image
boundary
value
Prior art date
Application number
PCT/CN2011/081964
Other languages
English (en)
French (fr)
Inventor
付廷杰
陈维强
李月高
刘韶
裴雷
Original Assignee
青岛海信网络科技股份有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 青岛海信网络科技股份有限公司 filed Critical 青岛海信网络科技股份有限公司
Publication of WO2013063820A1 publication Critical patent/WO2013063820A1/zh

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/60Type of objects
    • G06V20/62Text, e.g. of license plates, overlay texts or captions on TV images
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00Scenes; Scene-specific elements
    • G06V20/60Type of objects
    • G06V20/62Text, e.g. of license plates, overlay texts or captions on TV images
    • G06V20/625License plates

Definitions

  • the present invention relates to the field of communications technologies, and in particular, to a method and apparatus for positioning a license plate image.
  • BACKGROUND OF THE INVENTION Automatic license plate recognition can be applied to embedded high-definition electronic police and bayonet systems, and is a key link for intelligent transportation management.
  • the license plate recognition system is a highly intelligent integrated system based on computer image processing, pattern recognition and other technologies. The processing process includes license plate location, license plate character segmentation, and license plate character recognition.
  • the main task of license plate location is to locate the location of the license plate from the image of the captured vehicle and accurately extract the license plate for subsequent segmentation and identification of the license plate characters.
  • the accurate positioning of the vehicle image is the premise and basis for the correct recognition of the license plate characters, and is the key problem to be solved first by the license plate recognition technology.
  • the accuracy of license plate image positioning will directly affect the license plate character segmentation and recognition effect, which plays a vital role in the performance of the entire license plate recognition system.
  • Embodiments of the present invention provide a method and apparatus for positioning a license plate image, which can improve the accuracy of license plate location.
  • An embodiment of the present invention provides a method for locating a license plate image, including:
  • a license plate image boundary is determined by the first boundary, the second boundary, the third boundary, and the fourth boundary.
  • an embodiment of the present invention provides a device for locating a license plate image, including:
  • a first image acquisition module configured to acquire a color binarized image of the license plate image
  • a color projection acquisition module configured to use the number of pixels having color in each pixel row in the color binarized image as a color horizontal projection value of the pixel row, and the number of pixels having a color in each pixel column as the pixel column Vertical projection value of color;
  • a first boundary determining module configured to determine, according to the color horizontal projection value, a pixel row corresponding to a color projection height; and when the difference between the color projection height and the standard height is less than a first boundary threshold, the color projection a height corresponding initial pixel row and a ending pixel row as a first boundary and a second boundary of the license plate image;
  • a second boundary determining module configured to determine, according to the vertical projection value of the color, a pixel row corresponding to a color projection width; when the difference between the color projection width and the standard width is less than a second boundary threshold, the color projection width is Corresponding start pixel column and end pixel column as the third boundary and the fourth boundary of the license plate image;
  • a license plate positioning module configured to determine a license plate image boundary by the first boundary, the second boundary, the third boundary, and the fourth boundary.
  • Embodiments of the present invention provide a method and apparatus for locating a license plate image for acquiring a color binarized image of a license plate image; acquiring a color horizontal projection value of each pixel row of the color binarized image and each pixel column Color vertical projection value; according to the color horizontal projection value, obtaining a color projection height; when the difference between the color projection height and the standard height is less than the first boundary threshold, determining a starting pixel of the color horizontal projection to act on the license plate a first boundary of the image, a termination of the color horizontal projection, a second boundary of the license plate image; obtaining a color projection width according to the vertical projection value of the color; when the difference between the color projection width and the standard width is less than the second boundary
  • the starting pixel column of the vertical projection of the color is determined to be the third boundary of the license plate image, and the termination column of the vertical projection of the color is the fourth boundary of the license plate image.
  • FIG. 1 is a schematic flow chart of a method for positioning a license plate image according to an embodiment of the present invention
  • FIG. 2 is a schematic flow chart of a method for positioning a license plate image according to another embodiment of the present invention.
  • 3 is a schematic flowchart of a method for determining a maximum interval position according to an embodiment of the present invention
  • 4 is a schematic flowchart of a method for acquiring a grayscale binarized image according to an embodiment of the present invention
  • FIG. 5 is a schematic diagram of an apparatus for positioning a license plate image according to another embodiment of the present invention.
  • DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The main implementation principles, specific implementation manners, and the corresponding beneficial effects that can be achieved by the technical solutions of the embodiments of the present invention are described in detail below with reference to the accompanying drawings.
  • the embodiment of the present invention provides a method for locating a license plate image. As shown in FIG. 1, the method includes the following steps:
  • Step 101 Obtain a color binarized image of the license plate image
  • Step 102 The number of pixels having color in each pixel row in the color binarized image is used as a color horizontal projection value of the pixel row, and the number of pixels having color in each pixel column is used as a color vertical projection value of the pixel column; For example, each pixel row has 500 pixels, and in the first row, there are 200 pixels with color, and the color horizontal projection value of the row is 200.
  • Step 103 Determine, according to the color horizontal projection value, a pixel row corresponding to the color projection height; when the difference between the color projection height and the standard height is less than the first boundary threshold, the starting pixel row and the ending pixel row corresponding to the color projection height As a first boundary and a second boundary of the license plate image;
  • the color horizontal projection values of each row are respectively compared with the horizontal threshold; the pixel rows larger than the horizontal threshold are obtained; and in the pixel rows larger than the horizontal threshold, the number of consecutive rows having the largest number of consecutive rows is taken as the color projection height.
  • the width, height and position coordinates of the license plate image in at least two video images need to be counted; the correspondence between the width, the height and the position coordinates is established; and the standard width and the standard face of the current license plate image are determined according to the coordinates of the current license plate image.
  • the first boundary threshold may be one-sixth of the standard height, and the horizontal threshold may be one-half of the standard width, and the specific value may be set according to actual conditions.
  • the gray level projection value of each pixel row of the vehicle image is acquired according to the gray binarized image of the vehicle image; wherein, the grayscale image of the vehicle image
  • the number of pixels representing the license plate character in each pixel row is used as the gray level projection value of the pixel row; in the upper half of the license plate image, the grayscale horizontal projection value is less than 7 from the top to the bottom of the license plate image.
  • the last pixel row is the first boundary of the license plate image; in the lower half of the license plate image, the first pixel row having the projection number less than 7 is obtained from the top down as the second boundary of the license plate image.
  • the characters on the license plate are generally 7 digits. Therefore, when the boundary is judged here, 7 is used as the threshold.
  • Step 104 Determine, according to the vertical projection value of the color, a pixel column corresponding to the color projection width; when the difference between the color projection width and the standard width is less than the second boundary threshold, the starting pixel column and the ending pixel corresponding to the color projection width
  • the column serves as a third boundary and a fourth boundary of the license plate image
  • the vertical projection values of the colors of each column are respectively compared with the vertical threshold; the pixel columns larger than the vertical threshold are obtained; and in the pixel columns larger than the vertical threshold, the number of consecutive columns having the largest number of consecutive columns is taken as the color projection width.
  • the second boundary threshold may be one-eighth of the standard width, and the vertical threshold may be one-half of the standard height, and the specific value may be set according to actual conditions.
  • the maximum interval position is determined within the vehicle image when the difference between the color projection width and the standard width is not less than the second boundary threshold; and the third and fourth boundaries of the vehicle image are determined based on the maximum spacing position. Specifically, when determining the maximum interval position in the vehicle image, selecting one pixel column of the consecutive consecutive largest number of consecutive pixel columns whose gray vertical projection value is smaller than the interval threshold is the maximum interval position; wherein, the grayscale binarization of the vehicle image In the image, the number of pixels representing the license plate character in each pixel column is used as the gray vertical projection value of the pixel column.
  • the above interval threshold can be five-sixths of the standard height.
  • the maximum interval position is compared with the standard maximum interval position to obtain the difference; the third image of the vehicle image is determined according to the relationship between the standard maximum interval position and the standard width, and the difference The boundary and the fourth boundary.
  • the standard width is 100 pixels
  • the standard maximum interval is the 25th pixel from left to right
  • the standard maximum interval is 25 pixels from the left boundary (third boundary)
  • the current maximum interval position is 23 pixels from left to right, the difference between the two is 2 unit pixels, then the current maximum interval position is 25 pixels to the left of the starting point.
  • the pixel column where the pixel is terminated is the third boundary of the current license plate image; after the current maximum interval position is 75 pixels to the right of the starting point, the pixel column where the pixel is terminated is the fourth boundary of the current license plate image.
  • the third boundary and the fourth boundary of the vehicle image can still be determined by the above method.
  • the maximum interval position cannot be determined, or the boundary direction of the license plate image cannot be determined because there are few characters, the above method has a certain error.
  • the maximum interval position is determined in the left half of the license plate image; if the maximum interval position is in the right half of the left half of the vehicle image, the pixel column at the maximum interval position is the start Column, extending to the left side by a quarter of the standard width, extending the end of the pixel column as the third boundary; extending to the right side of the standard width of three quarters, extending the end of the pixel column as the fourth boundary;
  • the spacing position is in the left half of the left half of the vehicle image, and the pixel column in the maximum spacing position is the starting column, extending to the left side by one-eighth of the standard width, and the extending end point of the pixel column is the third boundary;
  • the right side extends seven-eighths of the standard width, and the pixel at the end of the extension is listed as the fourth boundary.
  • the maximum interval position may be between the first character and the second character (the left half in the left half of the vehicle image) in the actual license plate, or may be located between the second character and the third character (in the left half of the vehicle image)
  • the maximum interval position is determined in the left half of the vehicle image, and the pixel column whose gray vertical projection value is greater than the interval threshold is selected as the maximum interval position;
  • the gray-scale binarized image of the vehicle image the number of pixels representing the license plate character in each pixel column is used as the gray vertical projection value of the pixel column.
  • Obtaining a gray-scale binarized image of the vehicle image comprising: acquiring a gray histogram of the license plate image, and calculating a gray mean value of each pixel row of the license plate image; dividing the license plate image into at least four image regions vertically, and counting Gray mean value of each image area; comparing minimum gray mean and maximum gray mean in at least four image areas; if the difference between the two is less than a quarter of the minimum gray mean, then in the license plate image
  • the remaining image regions except the image region with the largest gray mean value are gray-scale compensated, and the binarization threshold is selected in the compensated gray histogram; the vehicle image is obtained according to the binarization threshold and the compensated license plate image.
  • Gray-scale binarized image if the difference between the two is not less than a quarter of the minimum gray mean, the binarization threshold is selected in the original gray histogram; according to the binarization threshold and the original license plate image Obtain a grayscale binarized image of the vehicle image.
  • the gradation value of the region with the smaller gradation mean value is increased to a predetermined value, so that the gradation mean value of the entire license plate is equalized, and the difference in gradation value due to reflection or the like is avoided.
  • the predetermined value may be an empirical value or a set value actually required.
  • the binarization threshold is selected in the compensated gray histogram or the original gray histogram, including: in the compensated gray histogram or the original gray histogram, the grayscale of the license plate character is smaller than the license plate
  • the ratio of the accumulated value to the total number of pixels is 0.3, the pixel obtained last time corresponds to a pixel point.
  • the gray value is used as the first gray value; when the ratio of the accumulated value to the total number of pixels is 0.4, the gray value corresponding to one pixel obtained last time is used as the second gray value; or the gray after compensation
  • the gray level of the license plate character is greater than the gray level of the license plate background, and the number of pixels corresponding to the accumulated gray value is from the right side of the gray histogram to the left side, when the accumulated value
  • the ratio of the total number of pixels is 0.3
  • the gray value corresponding to the last pixel obtained at this time is taken as the first gray value; when the ratio of the accumulated value to the total number of pixels is 0.4, the last time is obtained.
  • a gray value corresponding to one pixel as a second gray value a gray value between the first gray value and the second gray value, from small to large or large to small Arranging sequentially, and dividing into at least six data regions; using a minimum gray value in a data region having the smallest number of pixels among the at least six data regions as a binarization threshold.
  • Step 105 Determine a license plate image boundary from the first boundary, the second boundary, the third boundary, and the fourth boundary.
  • the method for positioning the license plate image obtained by the embodiment of the present invention obtains the standard width and standard through statistics by comprehensively considering the color information and the projection information, and combining the characteristics that the width and height of the license plate image are substantially the same. Height, thus making the license plate more precise positioning. Further, by performing grayscale binarization on the license plate image, the character features of the license plate image are effectively highlighted, and the maximum spacer position between the license plate characters is combined, and the license plate is more accurately positioned.
  • Step 201 preset a standard width and a standard height of the license plate image
  • calculating a width, a height, and a position coordinate of the license plate image in at least two video images establishing a correspondence relationship between the width, the height, and the position coordinate; determining a standard width of the current license plate image according to a coordinate of the current license plate image Standard height.
  • Step 202 Obtain a color projection height and a color projection width of a color binarized image of the current license plate image. Specifically, compare a color horizontal projection value of each row with a horizontal threshold; and obtain a pixel row greater than a horizontal threshold; In the pixel row of the horizontal threshold, the number of consecutive rows having the largest number of consecutive rows is taken as the color projection height.
  • the standard width is 1000 unit pixels
  • the horizontal threshold is 500 unit pixels
  • the number of pixels having the color of the first line of the current license plate image is 600, that is, the color horizontal projection value is 600 unit pixels
  • the projection value is 400 unit pixels
  • the color horizontal projection value of the third line is 700 unit pixels
  • the color horizontal projection value of the fourth line is 800 unit pixels
  • the color horizontal projection value of the fifth line is 850 unit pixels
  • the horizontal projection value is 300 unit pixels
  • the color horizontal projection values of the seventh to twentieth lines are all greater than 500 unit pixels
  • the color projection height is 14 unit pixels.
  • the vertical projection values of the colors of each column are respectively compared with the vertical threshold; the pixel columns larger than the vertical threshold are obtained; and in the pixel columns larger than the vertical threshold, the number of consecutive columns having the largest number of consecutive columns is taken as the color projection width.
  • Step 203 the difference between the color projection height and the standard height is compared with the first boundary threshold, if not, step 204 is performed; otherwise, step 205 is performed;
  • Step 204 Determine a first boundary of the color pixel horizontal projection of the license plate image, and a second boundary of the color behavior of the color horizontal projection; continue to perform step 207.
  • Step 205 Acquire a gray level projection value of the gray binarized image of the vehicle image.
  • Step 206 Compare the gray level horizontal projection value of each pixel row with 7 , and obtain, in the upper half of the license plate image, the last pixel row whose gray level horizontal projection value is less than 7 from the top to the bottom as the license plate image. A boundary; in the lower half of the license plate image, the first pixel row having a projection number less than 7 is obtained from the top down as the second boundary of the license plate image.
  • Step 207 the difference between the color projection width and the standard width is compared with the second boundary threshold; if less, step 208 is performed, otherwise step 209 is performed;
  • Step 208 Determine that the starting pixel column of the vertical projection of the color is the third boundary of the license plate image, and the termination column of the vertical projection of the color is the fourth boundary of the license plate image.
  • Step 209 Determine a maximum interval position of the image of the vehicle
  • Step 210 Determine a third boundary and a fourth boundary of the license plate image according to the maximum interval position. Specifically, if the maximum interval position is in the right half (between the second character and the third character) in the left half of the vehicle image, the pixel column in the maximum interval position is the starting column, and the standard width is extended to the left side.
  • the end point of the extended pixel is listed as the third boundary; the right side extends to three-quarters of the standard width, and the end point of the extended end is the fourth boundary; if the maximum interval position is in the left half of the vehicle image Half (between the first character and the second character), the pixel column in the maximum interval position is the starting column, extending to the left side by one-eighth of the standard width, and the extending end point of the pixel column is the third boundary; Extends the standard width by seven-eighths to the right, and the pixel at the end of the extension is listed as the fourth boundary. It is also possible to use the other correlation between the maximum interval position and the boundary to determine the boundary of the license plate image.
  • Step 211 Determine a license plate image boundary from the first boundary, the second boundary, the third boundary, and the fourth boundary Specifically, as shown in FIG. 3, when determining the maximum interval position in the foregoing step 209, the following steps may be performed: Step 301: Acquire a grayscale binarized image of the current license plate image;
  • Step 302 Obtain a gray vertical projection value of each pixel column. Specifically, the number of pixel points representing the license plate character in each pixel column is used as the gray vertical projection value of the pixel column. In the above step 205, the grayscale horizontal projection value is the number of pixels representing the license plate character in each pixel row.
  • Step 303 Compare a gray vertical projection value of each pixel column with an interval threshold; the interval threshold may be five-sixths of a standard height, or greater than five-sixths of a standard height.
  • Step 304 Select one pixel column of the consecutive consecutive largest number of consecutive pixel columns whose gray vertical projection value is smaller than the interval threshold as the maximum interval position.
  • Step 401 Obtain a gray histogram of the license plate image, and calculate a gray average value of each pixel row of the license plate image.
  • Step 402 vertically divide the license plate image into at least four image regions, and calculate a gray average value of each image region.
  • Step 403 Compare a minimum gray mean value and a maximum gray average value in at least four image regions; if the difference between the two is less than a quarter of the minimum gray mean value, perform step 404; otherwise, perform step 407. ;
  • Step 404 performing gray level compensation on the remaining image areas except the image area with the largest gray level mean value in the license plate image, and obtaining the compensated gray level histogram;
  • Step 405 Select a binarization threshold in the compensated gray histogram
  • Step 406 Acquire a gray binarized image of the vehicle image according to the binarization threshold and the compensated license plate image.
  • Step 407 Select a binarization threshold in the original gray histogram;
  • Step 408 Acquire a gray binarized image of the vehicle image according to the binarization threshold and the original grayscale histogram.
  • the binarization threshold is selected in the above steps 405 and 407, in the compensated gray histogram or the original gray histogram, the gray level of the license plate character is smaller than the gray level of the license plate background, from the gray histogram left The number of pixels corresponding to the gray value is accumulated from the lateral right side.
  • the gray value corresponding to the last pixel obtained at this time is taken as the first gray value;
  • the gray value corresponding to one pixel obtained last time is taken as the second gray value;
  • the gray level of the license plate character is greater than the gray level of the license plate background, and the number of pixels corresponding to the accumulated gray value is from the right side of the gray histogram to the left side.
  • the ratio of the accumulated value to the total number of pixels is 0.3, the gray value corresponding to the last pixel obtained at this time is taken as the first gray value; when the ratio of the accumulated value to the total number of pixels is 0.4, this is The gray value corresponding to one pixel obtained last time is used as the second gray value; or
  • a minimum gray value in a data region having the smallest number of pixels among the at least six data regions is used as a binarization threshold.
  • an embodiment of the present invention further provides a device for locating a license plate image, as shown in FIG. 5, including:
  • the first image obtaining module 501 is configured to obtain a color binarized image of the license plate image
  • the color projection acquisition module 502 is configured to use the number of pixels having color in each pixel row of the color binarized image as a color horizontal projection value of the pixel row, and the number of pixels having a color in each pixel column as the pixel.
  • a first boundary determining module 503 configured to determine, according to the color horizontal projection value, a pixel row corresponding to a color projection height; when the difference between the color projection height and the standard height is less than a first boundary threshold, the color is a starting pixel row and a ending pixel row corresponding to the projection height as a first boundary and a second boundary of the license plate image;
  • a second boundary determining module 504 configured to determine, according to the vertical projection value of the color, a pixel row corresponding to the color projection width; when the difference between the color projection width and the standard width is less than a second boundary threshold, the color projection a starting pixel column and a ending pixel column corresponding to the width as a third boundary and a fourth boundary of the license plate image;
  • a license plate location module 505 is configured to determine a license plate image boundary from the first boundary, the second boundary, the third boundary, and the fourth boundary.
  • the device further includes: a standard determining module 506, configured to count the average of the width and height of the license plate image in the at least two video images; the width average is used as the standard width, and the height average is used as the standard height.
  • a standard determining module 506 configured to count the average of the width and height of the license plate image in the at least two video images; the width average is used as the standard width, and the height average is used as the standard height.
  • the first boundary determining module 503 determines, according to the color horizontal projection value, a pixel row corresponding to the color projection height, specifically for comparing the color horizontal projection value of each row with a horizontal threshold; a pixel row of a horizontal threshold; in the pixel row larger than the horizontal threshold, the number of consecutive rows having the largest number of consecutive rows is taken as the color projection height.
  • the second boundary determining module 504 is configured to compare the vertical vertical projection values of each column with the vertical threshold according to the vertical projection value of the color to determine the pixel column corresponding to the color projection width; a pixel column of a vertical threshold; in the pixel column larger than the vertical threshold, the number of consecutive columns having the largest number of consecutive columns is taken as a color projection width.
  • the above device further comprises:
  • a third boundary determining module 507 configured to acquire, when the difference between the color projection height and the standard height is not less than the first boundary threshold, the gray of each pixel row of the vehicle image according to the gray-scale binarized image of the vehicle image a horizontal projection value; wherein, in the gray-scale binarized image of the vehicle image, the number of pixels representing the license plate character in each pixel row a grayscale horizontal projection value as the pixel row; in the upper half of the license plate image, obtaining a last pixel row whose grayscale horizontal projection value is less than 7 from the top to the bottom, as a first boundary of the license plate image; In the lower half of the license plate image, the first pixel row whose projection number is less than 7 is obtained from the top down as the second boundary of the license plate image.
  • the above device further comprises:
  • a fourth boundary determining module 508 configured to determine a maximum interval position within the vehicle image when a difference between the color projection width and a standard width is not less than a second boundary threshold; determining the vehicle according to the maximum interval position The third and fourth boundaries of the image.
  • the fourth boundary determining module 508 is specifically configured to: when the license plate image has a maximum interval position, the maximum The spacing position is compared with a standard maximum spacing position to obtain a difference; and a third boundary and a fourth boundary of the vehicle image are determined according to a relationship between the standard maximum spacing position and a standard width, and the difference.
  • the above device further comprises:
  • the interval bit determining module 509 is configured to select one pixel column of the continuous number of consecutive pixel columns whose gray vertical projection value is smaller than the interval threshold as the maximum interval position; wherein, the gray image binarized image of the vehicle image The number of pixels representing the license plate character in each pixel column is used as the gray vertical projection value of the pixel column.
  • the above device further comprises:
  • a second image acquisition module 510 configured to acquire a gray histogram of the license plate image, and calculate a gray average value of each pixel row of the license plate image; and vertically divide the license plate image into at least four image regions. Counting a gray level mean of each image region; comparing a minimum gray mean value and a maximum gray average value among the at least four image regions; if the difference between the two is less than a quarter of the minimum gray mean value, And performing gradation compensation on the remaining image regions except the image region with the largest gray mean value in the license plate image, and selecting a binarization threshold in the compensated gray histogram; according to the binarization threshold and compensation a subsequent license plate image, obtaining a gray binarized image of the vehicle image; if the difference between the two is not less than a quarter of the minimum gray mean value, selecting a binary value in the original gray histogram And thresholding the grayscale binarized image of the vehicle image according to the binarization threshold and the original grayscale histogram.
  • the above device further comprises:
  • the binarization threshold determination module 511 is configured to compensate the grayscale histogram or the original grayscale histogram, the gray level of the license plate character is smaller than the gray level of the license plate background, from the left side to the right side of the grayscale histogram Starting to accumulate the number of pixels corresponding to the gray value, when the ratio of the accumulated value to the total number of pixels is 0.3, the gray value corresponding to the last pixel obtained at this time is taken as the first gray value; when the accumulated value and the total value When the ratio of the number of pixels is 0.4, the gray value corresponding to one pixel obtained at this time is taken as the second gray value; or the gray of the license plate is grayed in the compensated gray histogram or the original gray histogram.
  • the degree is greater than the gray level of the license plate background.
  • the number of pixels corresponding to the cumulative gray value is from the right side of the gray histogram to the left side. When the ratio of the accumulated value to the total number of pixels is 0.3, the last acquired value will be obtained.
  • the gray value corresponding to one pixel is used as the first gray value; when the ratio of the accumulated value to the total number of pixels is 0.4, the last pixel obtained at this time is Corresponding gray value is used as the second gray value; the gray value between the first gray value and the second gray value is arranged in the order of small to large or large to small, And dividing into at least six data regions; the minimum gray value in the data region having the least pixel points among the at least six data regions is used as a binarization threshold.
  • the method and device for locating the license plate image obtained by the embodiment of the present invention obtains the standard width through statistics by comprehensively considering the color information and the projection information, and combining the characteristics that the width and height of the license plate image are substantially the same. And the standard height, which allows for a more precise positioning of the license plate. Further, by performing grayscale binarization on the license plate image, the character features of the license plate image are effectively highlighted, and the maximum spacer position between the license plate characters is combined to more accurately position the license plate.
  • embodiments of the present invention can be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or a combination of software and hardware. Moreover, the present invention can be embodied in the form of a computer program product embodied on one or more computer-usable storage interfaces (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.
  • computer-usable storage interfaces including but not limited to disk storage, CD-ROM, optical storage, etc.
  • the computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture comprising the instruction device.
  • the apparatus implements the functions specified in one or more blocks of a flow or a flow and/or block diagram of the flowchart.
  • These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device.
  • the instructions provide steps for implementing the functions specified in one or more of the flow or in a block or blocks of a flow diagram.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Character Input (AREA)
  • Image Analysis (AREA)
  • Traffic Control Systems (AREA)

Abstract

公开了一种车牌图像定位的方法和装置。该方法包括:根据颜色水平投影值,确定出颜色投影高度对应的像素行;当颜色投影高度与标准高度的差值小于第一边界阈值时,将颜色投影高度对应的起始像素行和终止像素行作为车牌图像的第一边界和第二边界;根据颜色垂直投影值,确定出颜色投影宽度对应的像素列;当颜色投影宽度与标准宽度的差值小于第二边界阈值时,将颜色投影宽度对应的起始像素列和终止像素列作为车牌图像的第三边界和第四边界。该车牌图像定位的方法和装置,可以准确进行车牌定位。

Description

一种车牌图像定位的方法和装置 本申请要求在 2011年 11月 1日提交中国专利局、申请号为 201110339877.X、发明名称为"一 种车牌图像定位的方法和装置"的中国专利申请的优先权, 其全部内容通过引用结合在本申请 中。 技术领域 本发明涉及通信领域技术, 尤其涉及一种车牌图像定位的方法和装置。 背景技术 车牌自动识别可以应用到嵌入式高清电子警察和卡口等系统中, 是实现交通管理智能 化的关键环节。 车牌识别系统是一个基于计算机图像处理、 模式识别等技术的高度智能化 的综合集成系统, 其处理流程包括车牌定位、 车牌字符分割、 车牌字符识别等。
车牌定位的主要任务是从拍摄的车辆图像中定位车牌所在的位置, 并把车牌准确的提 取出来, 供后续的车牌字符分割和识别使用。 车辆图像的准确定位是车牌字符正确识别的 前提和基础 , 是车牌识别技术首先要解决的关键问题。 车牌图像定位的准确性将直接影响 车牌字符分割和识别效果, 对整个车牌识别系统的性能起着至关重要的作用。
目前, 在车牌图像的定位方法中仍然存在很多的难点问题, 例如图像背景复杂、 光照 不均匀、 天气变化等。 正是由于环境因素的复杂性, 现有技术中使用的方法准确率不高, 处理周期也较长, 总之还不具有一种完全有效的解决方法。 发明内容 本发明实施例提供了一种车牌图像定位的方法和装置, 可以提高车牌定位的准确性。 本发明实施例提供了一种车牌图像定位的方法, 包括:
获取车牌图像的颜色二值化图像;
将所述颜色二值化图像中每一像素行具有颜色的像素点数目作为本像素行的颜色水 平投影值, 每一像素列具有颜色的像素点数目作为本像素列的颜色垂直投影值;
根据所述颜色水平投影值, 确定出颜色投影高度对应的像素行; 当所述颜色投影高度 与标准高度的差值小于第一边界阈值时, 将所述颜色投影高度对应的起始像素行和终止像 素行作为所述车牌图像的第一边界和第二边界;
根据所述颜色垂直投影值, 确定出颜色投影宽度对应的像素列; 当所述颜色投影宽度 与标准宽度的差值小于第二边界阈值时, 将所述颜色投影宽度对应的起始像素列和终止像 素列作为所述车牌图像的第三边界和第四边界;
由所述第一边界、 第二边界、 第三边界和第四边界确定出车牌图像边界。
相应的, 本发明实施例提供了一种车牌图像定位的装置, 包括:
第一图像获取模块, 用于获取车牌图像的颜色二值化图像;
颜色投影获取模块, 用于将所述颜色二值化图像中每一像素行具有颜色的像素点数目 作为本像素行的颜色水平投影值, 每一像素列具有颜色的像素点数目作为本像素列的颜色 垂直投影值;
第一边界确定模块, 用于根据所述颜色水平投影值, 确定出颜色投影高度对应的像素 行; 当所述颜色投影高度与标准高度的差值小于第一边界阈值时, 将所述颜色投影高度对 应的起始像素行和终止像素行作为所述车牌图像的第一边界和第二边界;
第二边界确定模块, 用于所述颜色垂直投影值, 确定出颜色投影宽度对应的像素行; 当所述颜色投影宽度与标准宽度的差值小于第二边界阈值时, 将所述颜色投影宽度对应的 起始像素列和终止像素列作为所述车牌图像的第三边界和第四边界;
车牌定位模块, 用于由所述第一边界、 第二边界、 第三边界和第四边界确定出车牌图 像边界。
本发明实施例提供了一种车牌图像定位的方法和装置, 用于获取车牌图像的颜色二值 化图像; 获取所述颜色二值化图像每一像素行的颜色水平投影值和每一像素列的颜色垂直 投影值; 根据所述颜色水平投影值, 获取颜色投影高度; 当所述颜色投影高度与标准高度 的差值小于第一边界阈值时, 确定颜色水平投影的起始像素行为所述车牌图像的第一边 界, 颜色水平投影的终止行为所述车牌图像的第二边界; 根据所述颜色垂直投影值, 获取 颜色投影宽度; 当所述颜色投影宽度与标准宽度的差值小于第二边界阈值时, 确定颜色垂 直投影的起始像素列为所述车牌图像的第三边界, 颜色垂直投影的终止列为所述车牌图像 的第四边界。 使用本发明实施例提供的车牌图像定位的方法和装置, 通过综合考虑颜色信 息和投影信息, 并结合车牌图像的宽度和高度基本一致的特点, 通过统计获得标准宽度和 标准高度, 由此对车牌进行较精确的定位。 进一步, 通过对车牌图像进行灰度二值化, 有 效的突出了车牌图像的字符特征, 并且结合了车牌字符间最大间隔符位置, 更为准确的对 车牌进行了定位。 附图说明 图 1为本发明实施例中车牌图像定位的方法流程示意图;
图 2为本发明另一实施例中车牌图像定位的方法流程示意图;
图 3为本发明实施例中确定最大间隔位置的方法流程示意图; 图 4为本发明实施例中获取灰度二值化图像的方法流程示意图;
图 5为本发明另一实施例中车牌图像定位的装置示意图。 具体实施方式 下面结合各个附图对本发明实施例技术方案的主要实现原理、 具体实施方式及其对应 能够达到的有益效果进行详细地阐述。
为了解决现有技术存在的问题, 本发明实施例提供了一种车牌图像定位的方法, 如图 1所示, 包括以下步骤:
步骤 101、 获取车牌图像的颜色二值化图像;
步骤 102、 将颜色二值化图像中每一像素行具有颜色的像素点数目作为本像素行的颜 色水平投影值,每一像素列具有颜色的像素点数目作为本像素列的颜色垂直投影值;例如, 每个像素行具有 500个像素点, 其中第某一行中具有颜色的像素点为 200个, 则该行的颜色 水平投影值为 200。
步骤 103、 根据颜色水平投影值, 确定出颜色投影高度对应的像素行; 当颜色投影高 度与标准高度的差值小于第一边界阈值时, 将颜色投影高度对应的起始像素行和终止像素 行作为车牌图像的第一边界和第二边界;
具体的, 将每一行的颜色水平投影值分别与水平阈值进行比较; 获取大于水平阈值的 像素行; 在大于水平阈值的像素行中, 将连续行数目最大的连续行数作为颜色投影高度。 其中, 需要统计至少两幅视频图像中车牌图像的宽度、 高度和位置坐标; 建立宽度、 高度 和位置坐标的对应关系; 根据当前车牌图像的坐标, 确定当前车牌图像的标准宽度和标准 面度。
上述第一边界阈值可以为标准高度的六分之一, 水平阈值可以为标准宽度的二分之 一 , 具体数值可以根据实际情况进行设定。
颜色投影高度与标准高度的差值不小于第一边界阈值时, 根据车辆图像的灰度二值化 图像获取车辆图像的每一像素行的灰度水平投影值;其中,车辆图像的灰度二值化图像中, 每一像素行中具代表车牌字符的像素点数目作为本像素行的灰度水平投影值; 在车牌图像 的上半部中, 从上向下获取灰度水平投影值小于 7的最后一个像素行, 作为车牌图像的第 一边界; 在车牌图像的下半部中, 从上向下获取投影数小于 7的第一个像素行, 作为车牌 图像的第二边界。 考虑到实际情况中, 车牌上的字符一般为 7位, 因此, 此处判断边界时 釆用 7作为阈值。
步骤 104、 根据颜色垂直投影值, 确定出颜色投影宽度对应的像素列; 当颜色投影宽 度与标准宽度的差值小于第二边界阈值时, 将颜色投影宽度对应的起始像素列和终止像素 列作为车牌图像的第三边界和第四边界;
具体的, 将每一列的颜色垂直投影值分别与垂直阈值进行比较; 获取大于垂直阈值的 像素列; 在大于垂直阈值的像素列中, 将连续列数目最大的连续列数作为颜色投影宽度。 上述第二边界阈值可以为标准宽度的八分之一, 垂直阈值可以为标准高度的二分之一, 具 体数值可以根据实际情况进行设定。
颜色投影宽度与标准宽度的差值不小于第二边界阈值时, 在车辆图像内确定最大间隔 位置; 根据最大间隔位置, 确定车辆图像的第三边界和第四边界。 具体的, 在车辆图像内 确定最大间隔位置时, 选取灰度垂直投影值小于间隔阈值的连续数目最大的连续像素列中 的一个像素列为最大间隔位置; 其中, 车辆图像的灰度二值化图像中, 每一像素列中代表 车牌字符的像素点数目作为本像素列的灰度垂直投影值。 上述间隔阈值可以为标准高度的 六分之五。 然后, 当车牌图像具有最大间隔位置时, 将最大间隔位置与标准最大间隔位置 进行比对, 获取差值; 根据标准最大间隔位置与标准宽度的关系、 以及该差值, 确定车辆 图像的第三边界和第四边界。 例如, 假设标准宽度为 100像素, 标准最大间隔位置为从左 到右第 25个像素点, 也就是该标准最大间隔位置距离左边界(第三边界) 25个单位像素, 距离右边界(第四边界) 75个单位像素。 当获取到的当前的最大间隔位置为从左到右第 23 个像素点, 则两者的差值为 2个单位像素, 那么以当前的最大间隔位置为起始点向左 25个 单位像素后, 终止像素点所在的像素列为当前车牌图像的第三边界; 以当前的最大间隔位 置为起始点向右 75个单位像素后, 终止像素点所在的像素列为当前车牌图像的第四边界。 其中, 当车牌图像不具有完整数目的字符, 但具有最大间隔位置时, 仍然可以利用上述方 式确定出车辆图像的第三边界和第四边界。 但是当车牌图像过于残缺, 无法确定最大间隔 位置或者由于字符较少无法确定车牌图像的边界方向时, 使用上述方式具有一定的误差。
或者当车牌图像具有完整数目的车牌字符时, 在车牌图像的左半部内确定最大间隔位 置; 若最大间隔位置在车辆图像左半部内的右半部, 则在最大间隔位置所在像素列为起始 列, 向左侧延伸标准宽度的四分之一, 延伸的终点所在像素列为第三边界; 向右侧延伸标 准宽度的四分之三, 延伸的终点所在像素列为第四边界; 若最大间隔位置在车辆图像左半 部内的左半部,则在最大间隔位置所在像素列为起始列,向左侧延伸标准宽度的八分之一, 延伸的终点所在像素列为第三边界; 向右侧延伸标准宽度的八分之七, 延伸的终点所在像 素列为第四边界。其中,最大间隔位置在实际车牌中可能位于第一字符和第二字符之间(车 辆图像左半部内的左半部) , 也可能位于第二字符和第三字符之间 (车辆图像左半部内的 右半部) , 基于这一规律, 本发明实施例提供的方法中在车辆图像的左半部内确定最大间 隔位置, 选取灰度垂直投影值大于间隔阈值的像素列为最大间隔位置; 其中, 车辆图像的 灰度二值化图像中, 每一像素列中代表车牌字符的像素点数目作为本像素列的灰度垂直投 影值。 获取车辆图像的灰度二值化图像时, 包括: 获取车牌图像的灰度直方图, 统计车牌图 像的每个像素行的灰度均值; 将车牌图像纵向均分为至少四个图像区域, 统计每个图像区 域的灰度均值; 将至少四个图像区域中最小灰度均值和最大灰度均值进行比较; 若两者的 差值小于最小灰度均值的四分之一 , 则对车牌图像中除灰度均值最大的图像区域外的其余 图像区域均进行灰度补偿, 在补偿后的灰度直方图中选取二值化阈值; 根据二值化阈值和 补偿后的车牌图像, 获取车辆图像的灰度二值化图像; 若两者的差值不小于最小灰度均值 的四分之一, 则在原始的灰度直方图中选取二值化阈值; 根据二值化阈值和原始的车牌图 像, 获取车辆图像的灰度二值化图像。 进行灰度补偿时, 将灰度均值较小的区域的灰度值 增加到预定值, 使得整个车牌的灰度均值得到均衡, 避免由于反光等原因造成的灰度值差 异。 而且, 该预定值可以为经验值, 也可以为实际需要的设定值。
其中, 在补偿后的灰度直方图或原始的灰度直方图中选取二值化阈值, 包括: 补偿后 的灰度直方图或原始的灰度直方图中, 车牌字符的灰度级小于车牌背景的灰度级, 从灰度 直方图左侧向右侧开始累积灰度值对应的像素点数目, 当累加值与总像素点数目比值为 0.3 时, 将此时最后获取的一个像素点对应的灰度值作为第一灰度值; 当累加值与总像素点数 目比值为 0.4时, 将此时最后获取的一个像素点对应的灰度值作为第二灰度值; 或者补偿后 的灰度直方图或原始的灰度直方图中, 车牌字符的灰度级大于车牌背景的灰度级, 从灰度 直方图右侧向左侧开始累积灰度值对应的像素点数目, 当累加值与总像素点数目比值为 0.3 时, 将此时最后获取的一个像素点对应的灰度值作为第一灰度值; 当累加值与总像素点数 目比值为 0.4时, 将此时最后获取的一个像素点对应的灰度值作为第二灰度值; 将位于所述 第一灰度值和所述第二灰度值之间的灰度值, 按照从小到大或从大到小的顺序进行排列, 并均分为至少六个数据区域; 将所述至少六个数据区域中像素点最少的数据区域中的最小 灰度值作为二值化阈值。
步骤 105、 由第一边界、 第二边界、 第三边界和第四边界确定出车牌图像边界。
通过上述描述, 可以看出, 使用本发明实施例提供的车牌图像定位的方法, 通过综合 考虑颜色信息和投影信息, 并结合车牌图像的宽度和高度基本一致的特点, 通过统计获得 标准宽度和标准高度, 由此对车牌进行较精确的定位。 进一步, 通过对车牌图像进行灰度 二值化, 有效的突出了车牌图像的字符特征, 并且结合了车牌字符间最大间隔符位置, 更 为准确的对车牌进行了定位。
下面通过具体实施例对本发明提供的方法进行详细说明, 如图 2所示, 具体包括以下 步骤:
步骤 201、 预先设置车牌图像的标准宽度和标准高度;
具体的, 统计至少两幅视频图像中车牌图像的宽度、 高度和位置坐标; 建立所述宽度、 高度和位置坐标的对应关系; 根据当前车牌图像的坐标, 确定当前车牌图像的标准宽度和 标准高度。
步骤 202、 获取当前车牌图像的颜色二值化图像的颜色投影高度和颜色投影宽度; 具 体的,将每一行的颜色水平投影值分别与水平阈值进行比较;获取大于水平阈值的像素行; 在大于水平阈值的像素行中, 将连续行数目最大的连续行数作为颜色投影高度。 例如, 该 标准宽度为 1000单位像素, 水平阈值为 500单位像素, 该当前车牌图像的第一行具有颜色 的像素点数目为 600 , 即颜色水平投影值为 600单位像素; 第二行的颜色水平投影值为 400 单位像素; 第三行的颜色水平投影值为 700单位像素; 第四行的颜色水平投影值为 800单位 像素; 第五行的颜色水平投影值为 850单位像素,第六行的颜色水平投影值为 300单位像素; 第七行至第二十行的颜色水平投影值均大于 500单位像素, 则颜色投影高度为 14单位像素。
将每一列的颜色垂直投影值分别与垂直阈值进行比较; 获取大于垂直阈值的像素列; 在大于垂直阈值的像素列中, 将连续列数目最大的连续列数作为颜色投影宽度。
步骤 203、 将颜色投影高度和标准高度的差值与第一边界阈值进行比较, 若小于, 则 执行步骤 204; 否则, 执行步骤 205;
步骤 204、 确定颜色水平投影的起始像素行为车牌图像的第一边界, 颜色水平投影的 终止行为车牌图像的第二边界; 继续执行步骤 207。
步骤 205、 获取该车辆图像的灰度二值化图像的灰度水平投影值;
步骤 206、 将每个像素行的灰度水平投影值与 7进行比较, 在车牌图像的上半部中, 从 上向下获取灰度水平投影值小于 7的最后一个像素行作为车牌图像的第一边界; 在车牌图 像的下半部中, 从上向下获取投影数小于 7的第一个像素行, 作为车牌图像的第二边界。
步骤 207、 将颜色投影宽度和标准宽度的差值与第二边界阈值进行比较; 若小于, 则 执行步骤 208 , 否则执行步骤 209;
步骤 208、 确定颜色垂直投影的起始像素列为车牌图像的第三边界, 颜色垂直投影的 终止列为车牌图像的第四边界。
步骤 209、 确定车辆图像的最大间隔位置;
步骤 210、 根据最大间隔位置确定车牌图像的第三边界和第四边界。 具体的, 若最大 间隔位置在车辆图像左半部内的右半部 (第二字符和第三字符之间) , 则在最大间隔位置 所在像素列为起始列, 向左侧延伸标准宽度的四分之一, 延伸的终点所在像素列为第三边 界; 向右侧延伸标准宽度的四分之三, 延伸的终点所在像素列为第四边界; 若最大间隔位 置在车辆图像左半部内的左半部 (第一字符和第二字符之间) , 则在最大间隔位置所在像 素列为起始列, 向左侧延伸标准宽度的八分之一, 延伸的终点所在像素列为第三边界; 向 右侧延伸标准宽度的八分之七, 延伸的终点所在像素列为第四边界。 还可以利用该最大间 隔位置与边界的其他相关关系, 确定出车牌图像的边界。
步骤 211、 由第一边界、 第二边界、 第三边界和第四边界确定出车牌图像边界 具体, 如图 3所示, 上述步骤 209中确定最大间隔位置时可以执行以下步骤: 步骤 301、 获取当前车牌图像的灰度二值化图像;
步骤 302、 获取每个像素列的灰度垂直投影值; 具体的, 每一像素列中代表车牌字符 的像素点数目作为本像素列的灰度垂直投影值。 上述步骤 205中, 灰度水平投影值为每一 像素行中代表车牌字符的像素点数目。
步骤 303、 将每个像素列的灰度垂直投影值均与间隔阈值进行比较; 该间隔阈值可以 为标准高度的六分之五, 或者大于标准高度的六分之五。
步骤 304、 选取灰度垂直投影值小于间隔阈值的连续数目最大的连续像素列的中的一 个像素列为最大间隔位置。
上述步骤 301中获取灰度二值化图像时, 如图 4所示, 执行以下步骤:
步骤 401、 获取车牌图像的灰度直方图, 统计车牌图像的每个像素行的灰度均值; 步骤 402、 将车牌图像纵向均分为至少四个图像区域, 统计每个图像区域的灰度均值; 步骤 403、 将至少四个图像区域中最小灰度均值和最大灰度均值进行比较; 若两者的 差值小于最小灰度均值的四分之一, 则执行步骤 404; 否则, 执行步骤 407;
步骤 404、 对车牌图像中除灰度均值最大的图像区域外的其余图像区域均进行灰度补 偿, 获得补偿后的灰度直方图;
步骤 405、 在补偿后的灰度直方图中选取二值化阈值;
步骤 406、 根据二值化阈值和补偿后的车牌图像, 获取车辆图像的灰度二值化图像。 步骤 407、 在原始的灰度直方图中选取二值化阈值;
步骤 408、 根据二值化阈值和原始的灰度直方图, 获取车辆图像的灰度二值化图像。 上述步骤 405和步骤 407中选取二值化阈值时, 补偿后的灰度直方图或原始的灰度直方 图中, 车牌字符的灰度级小于车牌背景的灰度级, 从灰度直方图左侧向右侧开始累积灰度 值对应的像素点数目, 当累加值与总像素点数目比值为 0.3时, 将此时最后获取的一个像素 点对应的灰度值作为第一灰度值; 当累加值与总像素点数目比值为 0.4时, 将此时最后获取 的一个像素点对应的灰度值作为第二灰度值; 或者
补偿后的灰度直方图或原始的灰度直方图中, 车牌字符的灰度级大于车牌背景的灰度 级, 从灰度直方图右侧向左侧开始累积灰度值对应的像素点数目, 当累加值与总像素点数 目比值为 0.3时, 将此时最后获取的一个像素点对应的灰度值作为第一灰度值; 当累加值与 总像素点数目比值为 0.4时, 将此时最后获取的一个像素点对应的灰度值作为第二灰度值; 或者
将位于所述第一灰度值和所述第二灰度值之间的灰度值 , 按照从小到大或从大到小的 顺序进行排列, 并均分为至少六个数据区域;
将所述至少六个数据区域中像素点最少的数据区域中的最小灰度值作为二值化阈值。 通过上述描述, 可以看出, 使用本发明实施例提供的车牌图像定位的方法, 通过综合 考虑颜色信息和投影信息, 并结合车牌图像的宽度和高度基本一致的特点, 通过统计获得 标准宽度和标准高度, 由此对车牌进行较精确的定位。 进一步, 通过对车牌图像进行灰度 二值化, 有效的突出了车牌图像的字符特征, 并且结合了车牌字符间最大间隔符位置, 更 为准确的对车牌进行了定位。
基于同一发明构思, 本发明实施例还提供了一种车牌图像定位的装置, 如图 5所示, 包括:
第一图像获取模块 501 , 用于获取车牌图像的颜色二值化图像;
颜色投影获取模块 502 , 用于将所述颜色二值化图像中每一像素行具有颜色的像素点 数目作为本像素行的颜色水平投影值, 每一像素列具有颜色的像素点数目作为本像素列的 颜色垂直投影值;
第一边界确定模块 503 , 用于根据所述颜色水平投影值, 确定出颜色投影高度对应的 像素行; 当所述颜色投影高度与标准高度的差值小于第一边界阈值时, 将所述颜色投影高 度对应的起始像素行和终止像素行作为所述车牌图像的第一边界和第二边界;
第二边界确定模块 504 , 用于所述颜色垂直投影值, 确定出颜色投影宽度对应的像素 行; 当所述颜色投影宽度与标准宽度的差值小于第二边界阈值时, 将所述颜色投影宽度对 应的起始像素列和终止像素列作为所述车牌图像的第三边界和第四边界;
车牌定位模块 505 , 用于由所述第一边界、 第二边界、 第三边界和第四边界确定出车 牌图像边界。
较佳的, 上述装置还包括: 标准确定模块 506 , 用于统计至少两幅视频图像中车牌图 像的宽度和高度的均值; 将宽度均值作为标准宽度, 将高度均值作为标准高度。
较佳的, 第一边界确定模块 503根据所述颜色水平投影值, 确定出颜色投影高度对应 的像素行时, 具体用于将每一行的颜色水平投影值分别与水平阈值进行比较; 获取大于所 述水平阈值的像素行; 在所述大于所述水平阈值的像素行中, 将连续行数目最大的连续行 数作为颜色投影高度。
较佳的, 第二边界确定模块 504根据所述颜色垂直投影值, 确定出颜色投影宽度对应 的像素列时, 具体用于将每一列的颜色垂直投影值分别与垂直阈值进行比较; 获取大于所 述垂直阈值的像素列; 在所述大于所述垂直阈值的像素列中, 将连续列数目最大的连续列 数作为颜色投影宽度。
较佳的, 上述装置还包括:
第三边界确定模块 507 , 用于所述颜色投影高度与标准高度的差值不小于第一边界阈 值时, 根据车辆图像的灰度二值化图像获取所述车辆图像的每一像素行的灰度水平投影 值; 其中, 所述车辆图像的灰度二值化图像中, 每一像素行中代表车牌字符的像素点数目 作为本像素行的灰度水平投影值; 在所述车牌图像的上半部中, 从上向下获取灰度水平投 影值小于 7的最后一个像素行, 作为车牌图像的第一边界; 在所述车牌图像的下半部中, 从上向下获取投影数小于 7的第一个像素行, 作为车牌图像的第二边界。
较佳的, 上述装置还包括:
第四边界确定模块 508 , 用于所述颜色投影宽度与标准宽度的差值不小于第二边界阈 值时, 在所述车辆图像内确定最大间隔位置; 根据所述最大间隔位置, 确定所述车辆图像 的第三边界和第四边界。
较佳的, 第四边界确定模块 508根据所述最大间隔位置, 确定所述车辆图像的第三边 界和第四边界时, 具体用于当所述车牌图像具有最大间隔位置时, 将所述最大间隔位置与 标准最大间隔位置进行比对, 获取差值; 根据所述标准最大间隔位置与标准宽度的关系、 以及所述差值, 确定所述车辆图像的第三边界和第四边界。
较佳的, 上述装置还包括:
间隔位确定模块 509 , 用于选取灰度垂直投影值小于间隔阈值的连续数目最大的连续 像素列的中的一个像素列为最大间隔位置; 其中, 所述车辆图像的灰度二值化图像中, 每 一像素列中代表车牌字符的像素点数目作为本像素列的灰度垂直投影值。
较佳的, 上述装置还包括:
第二图像获取模块 510 , 用于获取所述车牌图像的灰度直方图, 统计所述车牌图像的 每个像素行的灰度均值; 将所述车牌图像纵向均分为至少四个图像区域, 统计每个图像区 域的灰度均值; 将所述至少四个图像区域中最小灰度均值和最大灰度均值进行比较; 若两 者的差值小于所述最小灰度均值的四分之一, 则对所述车牌图像中除灰度均值最大的图像 区域外的其余图像区域均进行灰度补偿, 在补偿后的灰度直方图中选取二值化阈值; 根据 所述二值化阈值和补偿后的车牌图像, 获取所述车辆图像的灰度二值化图像; 若两者的差 值不小于所述最小灰度均值的四分之一, 则在原始的灰度直方图中选取二值化阈值; 根据 所述二值化阈值和原始的灰度直方图, 获取所述车辆图像的灰度二值化图像。
较佳的, 上述装置还包括:
二值化阈值确定模块 511 , 用于补偿后的灰度直方图或原始的灰度直方图中, 车牌字 符的灰度级小于车牌背景的灰度级, 从灰度直方图左侧向右侧开始累积灰度值对应的像素 点数目, 当累加值与总像素点数目比值为 0.3时, 将此时最后获取的一个像素点对应的灰度 值作为第一灰度值; 当累加值与总像素点数目比值为 0.4时, 将此时最后获取的一个像素点 对应的灰度值作为第二灰度值; 或者补偿后的灰度直方图或原始的灰度直方图中, 车牌字 符的灰度级大于车牌背景的灰度级, 从灰度直方图右侧向左侧开始累积灰度值对应的像素 点数目, 当累加值与总像素点数目比值为 0.3时, 将此时最后获取的一个像素点对应的灰度 值作为第一灰度值; 当累加值与总像素点数目比值为 0.4时, 将此时最后获取的一个像素点 对应的灰度值作为第二灰度值; 将位于所述第一灰度值和所述第二灰度值之间的灰度值 , 按照从小到大或从大到小的顺序进行排列, 并均分为至少六个数据区域; 将所述至少六个 数据区域中像素点最少的数据区域中的最小灰度值作为二值化阈值。
通过上述描述, 可以看出, 使用本发明实施例提供的车牌图像定位的方法和装置, 通 过综合考虑颜色信息和投影信息, 并结合车牌图像的宽度和高度基本一致的特点, 通过统 计获得标准宽度和标准高度, 由此对车牌进行较精确的定位。 进一步, 通过对车牌图像进 行灰度二值化, 有效的突出了车牌图像的字符特征, 并且结合了车牌字符间最大间隔符位 置, 更为准确的对车牌进行了定位。
本领域内的技术人员应明白, 本发明的实施例可提供为方法、 系统、 或计算机程序产 品。 因此, 本发明可釆用完全硬件实施例、 完全软件实施例、 或结合软件和硬件方面的实 施例的形式。 而且, 本发明可釆用在一个或多个其中包含有计算机可用程序代码的计算机 可用存储介盾 (包括但不限于磁盘存储器、 CD-ROM、 光学存储器等)上实施的计算机程 序产品的形式。
本发明是参照根据本发明实施例的方法、 设备(系统) 、 和计算机程序产品的流程图 和 /或方框图来描述的。 应理解可由计算机程序指令实现流程图和 /或方框图中的每一流 程和 /或方框、 以及流程图和 /或方框图中的流程和 /或方框的结合。 可提供这些计算机 程序指令到通用计算机、 专用计算机、 嵌入式处理机或其他可编程数据处理设备的处理器 以产生一个机器, 使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用 于实现在流程图一个流程或多个流程和 /或方框图一个方框或多个方框中指定的功能的 装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方 式工作的计算机可读存储器中, 使得存储在该计算机可读存储器中的指令产生包括指令装 置的制造品, 该指令装置实现在流程图一个流程或多个流程和 /或方框图一个方框或多个 方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上, 使得在计算机 或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理, 从而在计算机或其他 可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和 /或方框图一个 方框或多个方框中指定的功能的步骤。
尽管已描述了本发明的优选实施例, 但本领域内的技术人员一旦得知了基本创造性概 念, 则可对这些实施例作出另外的变更和修改。 所以, 所附权利要求意欲解释为包括优选 实施例以及落入本发明范围的所有变更和修改。
显然, 本领域的技术人员可以对本发明实施例进行各种改动和变型而不脱离本发明实 施例的精神和范围。 这样, 倘若本发明实施例的这些修改和变型属于本发明权利要求及其 等同技术的范围之内, 则本发明也意图包含这些改动和变型在内。

Claims

权 利 要 求
1、 一种车牌图像定位的方法, 其特征在于, 该方法包括:
获取车牌图像的颜色二值化图像;
将所述颜色二值化图像中每一像素行具有颜色的像素点数目作为本像素行的颜色水 平投影值, 每一像素列具有颜色的像素点数目作为本像素列的颜色垂直投影值;
根据所述颜色水平投影值, 确定出颜色投影高度对应的像素行; 当所述颜色投影高度 与标准高度的差值小于第一边界阈值时, 将所述颜色投影高度对应的起始像素行和终止像 素行作为所述车牌图像的第一边界和第二边界;
根据所述颜色垂直投影值, 确定出颜色投影宽度对应的像素列; 当所述颜色投影宽度 与标准宽度的差值小于第二边界阈值时, 将所述颜色投影宽度对应的起始像素列和终止像 素列作为所述车牌图像的第三边界和第四边界;
由所述第一边界、 第二边界、 第三边界和第四边界确定出车牌图像边界。
2、如权利要求 1所述的方法, 其特征在于, 所述获取车牌图像的颜色二值化图像之前, 还包括:
统计至少两幅视频图像中车牌图像的宽度、 高度和位置坐标;
建立所述宽度、 高度和位置坐标的对应关系;
根据当前车牌图像的坐标, 确定当前车牌图像的标准宽度和标准高度。
3、 如权利要求 1所述的方法, 其特征在于, 根据所述颜色水平投影值, 确定出颜色投 影高度对应的像素行, 包括:
将每一行的颜色水平投影值分别与水平阈值进行比较;
获取大于所述水平阈值的像素行;
在所述大于所述水平阈值的像素行中, 将连续行数目最大的连续行数作为颜色投影高 度。
4、 如权利要求 1所述的方法, 其特征在于, 所述根据所述颜色垂直投影值, 确定出颜 色投影宽度对应的像素列, 包括:
将每一列的颜色垂直投影值分别与垂直阈值进行比较;
获取大于所述垂直阈值的像素列;
在所述大于所述垂直阈值的像素列中, 将连续列数目最大的连续列数作为颜色投影宽 度。
5、 如权利要求 1所述的方法, 其特征在于, 所述颜色投影高度与标准高度的差值不小 于第一边界阈值时, 根据车辆图像的灰度二值化图像获取所述车辆图像的每一像素行的灰 度水平投影值; 其中, 所述车辆图像的灰度二值化图像中, 每一像素行中代表车牌字符的 像素点数目作为本像素行的灰度水平投影值;
在所述车牌图像的上半部中, 从上向下获取灰度水平投影值小于 7的最后一个像素行, 作为车牌图像的第一边界; 在所述车牌图像的下半部中, 从上向下获取投影数小于 7的第 一个像素行, 作为车牌图像的第二边界。
6、 如权利要求 1所述的方法, 其特征在于, 所述颜色投影宽度与标准宽度的差值不小 于第二边界阈值时, 在所述车辆图像内确定最大间隔位置;
根据所述最大间隔位置, 确定所述车辆图像的第三边界和第四边界。
7、 如权利要求 6所述的方法, 其特征在于, 在所述车辆图像内确定最大间隔位置, 包 括:
选取灰度垂直投影值小于间隔阈值的连续数目最大的连续像素列中的一个像素列为 最大间隔位置; 其中, 所述车辆图像的灰度二值化图像中, 每一像素列中代表车牌字符的 像素点数目作为本像素列的灰度垂直投影值。
8、 如权利要求 7所述的方法, 其特征在于, 根据所述最大间隔位置, 确定所述车辆图 像的第三边界和第四边界, 包括:
当所述车牌图像具有最大间隔位置时, 将所述最大间隔位置与标准最大间隔位置进行 比对, 获取差值;
根据所述标准最大间隔位置与标准宽度的关系、 以及所述差值, 确定所述车辆图像的 第三边界和第四边界。
9、 如权利要求 7所述的方法, 其特征在于, 获取所述车辆图像的灰度二值化图像时, 包括:
获取所述车牌图像的灰度直方图, 统计所述车牌图像的每个像素行的灰度均值; 将所述车牌图像纵向均分为至少四个图像区域, 统计每个图像区域的灰度均值; 将所述至少四个图像区域中最小灰度均值和最大灰度均值进行比较;
若两者的差值小于所述最小灰度均值的四分之一, 则对所述车牌图像中除灰度均值最 大的图像区域外的其余图像区域均进行灰度补偿, 在补偿后的灰度直方图中选取二值化阈 值; 根据所述二值化阈值和补偿后的车牌图像, 获取所述车辆图像的灰度二值化图像; 若两者的差值不小于所述最小灰度均值的四分之一, 则在原始的灰度直方图中选取二 值化阈值;根据所述二值化阈值和原始的车牌图像,获取所述车辆图像的灰度二值化图像。
10、 如权利要求 9所述的方法, 其特征在于, 在补偿后的灰度直方图或原始的灰度直 方图中选取二值化阈值, 包括:
补偿后的灰度直方图或原始的灰度直方图中, 车牌字符的灰度级小于车牌背景的灰度 级, 从灰度直方图左侧向右侧开始累积灰度值对应的像素点数目, 当累加值与总像素点数 目比值为 0.3时, 将此时最后获取的一个像素点对应的灰度值作为第一灰度值; 当累加值与 总像素点数目比值为 0.4时, 将此时最后获取的一个像素点对应的灰度值作为第二灰度值; 或者
补偿后的灰度直方图或原始的灰度直方图中, 车牌字符的灰度级大于车牌背景的灰度 级, 从灰度直方图右侧向左侧开始累积灰度值对应的像素点数目, 当累加值与总像素点数 目比值为 0.3时, 将此时最后获取的一个像素点对应的灰度值作为第一灰度值; 当累加值与 总像素点数目比值为 0.4时, 将此时最后获取的一个像素点对应的灰度值作为第二灰度值; 将位于所述第一灰度值和所述第二灰度值之间的灰度值 , 按照从小到大或从大到小的 顺序进行排列, 并均分为至少六个数据区域;
将所述至少六个数据区域中像素点最少的数据区域中的最小灰度值作为二值化阈值。
11、 一种车牌图像定位的装置, 其特征在于, 包括:
第一图像获取模块, 用于获取车牌图像的颜色二值化图像;
颜色投影获取模块, 用于将所述颜色二值化图像中每一像素行具有颜色的像素点数目 作为本像素行的颜色水平投影值, 每一像素列具有颜色的像素点数目作为本像素列的颜色 垂直投影值;
第一边界确定模块, 用于根据所述颜色水平投影值, 确定出颜色投影高度对应的像素 行; 当所述颜色投影高度与标准高度的差值小于第一边界阈值时, 将所述颜色投影高度对 应的起始像素行和终止像素行作为所述车牌图像的第一边界和第二边界;
第二边界确定模块, 用于所述颜色垂直投影值, 确定出颜色投影宽度对应的像素行; 当所述颜色投影宽度与标准宽度的差值小于第二边界阈值时, 将所述颜色投影宽度对应的 起始像素列和终止像素列作为所述车牌图像的第三边界和第四边界;
车牌定位模块, 用于由所述第一边界、 第二边界、 第三边界和第四边界确定出车牌图 像边界。
12、 如权利要求 11所述的装置, 其特征在于, 还包括: 标准确定模块, 用于统计至少 两幅视频图像中车牌图像的宽度、 高度和位置坐标; 建立所述宽度、 高度和位置坐标的对 应关系; 根据当前车牌图像的坐标, 确定当前车牌图像的标准宽度和标准高度。
13、 如权利要求 11所述的装置, 其特征在于, 所述第一边界确定模块根据所述颜色水 平投影值, 确定出颜色投影高度对应的像素行时, 具体用于将每一行的颜色水平投影值分 别与水平阈值进行比较; 获取大于所述水平阈值的像素行; 在所述大于所述水平阈值的像 素行中, 将连续行数目最大的连续行数作为颜色投影高度。
14、 如权利要求 11所述的装置, 其特征在于, 所述第二边界确定模块根据所述颜色垂 直投影值, 确定出颜色投影宽度对应的像素列时, 具体用于将每一列的颜色垂直投影值分 别与垂直阈值进行比较; 获取大于所述垂直阈值的像素列; 在所述大于所述垂直阈值的像 素列中, 将连续列数目最大的连续列数作为颜色投影宽度。
15、 如权利要求 11所述的装置, 其特征在于, 还包括:
第三边界确定模块, 用于所述颜色投影高度与标准高度的差值不小于第一边界阈值 时, 根据车辆图像的灰度二值化图像获取所述车辆图像的每一像素行的灰度水平投影值; 其中, 所述车辆图像的灰度二值化图像中, 每一像素行中代表车牌字符的像素点数目作为 本像素行的灰度水平投影值; 在所述车牌图像的上半部中, 从上向下获取灰度水平投影值 小于 7的最后一个像素行, 作为车牌图像的第一边界; 在所述车牌图像的下半部中, 从上 向下获取投影数小于 7的第一个像素行, 作为车牌图像的第二边界。
16、 如权利要求 11所述的装置, 其特征在于, 还包括:
第四边界确定模块, 用于所述颜色投影宽度与标准宽度的差值不小于第二边界阈值 时, 在所述车辆图像内确定最大间隔位置; 根据所述最大间隔位置, 确定所述车辆图像的 第三边界和第四边界。
17、 如权利要求 16所述的装置, 其特征在于, 所述第四边界确定模块根据所述最大间 隔位置, 确定所述车辆图像的第三边界和第四边界时, 具体用于当所述车牌图像具有最大 间隔位置时, 将所述最大间隔位置与标准最大间隔位置进行比对, 获取差值; 根据所述标 准最大间隔位置与标准宽度的关系、 以及所述差值, 确定所述车辆图像的第三边界和第四 边界。
18、 如权利要求 16所述的装置, 其特征在于, 还包括:
间隔位确定模块, 用于选取灰度垂直投影值小于间隔阈值的连续数目最大的连续像素 列的中的一个像素列为最大间隔位置; 其中, 所述车辆图像的灰度二值化图像中, 每一像 素列中代表车牌字符的像素点数目作为本像素列的灰度垂直投影值。
19、 如权利要求 18所述的装置, 其特征在于, 还包括:
第二图像获取模块, 用于获取所述车牌图像的灰度直方图, 统计所述车牌图像的每个 像素行的灰度均值; 将所述车牌图像纵向均分为至少四个图像区域, 统计每个图像区域的 灰度均值; 将所述至少四个图像区域中最小灰度均值和最大灰度均值进行比较; 若两者的 差值小于所述最小灰度均值的四分之一 , 则对所述车牌图像中除灰度均值最大的图像区域 外的其余图像区域均进行灰度补偿, 在补偿后的灰度直方图中选取二值化阈值; 根据所述 二值化阈值和补偿后的车牌图像, 获取所述车辆图像的灰度二值化图像; 若两者的差值不 小于所述最小灰度均值的四分之一, 则在原始的灰度直方图中选取二值化阈值; 根据所述 二值化阈值和原始的车牌图像, 获取所述车辆图像的灰度二值化图像。
20、 如权利要求 17所述的装置, 其特征在于, 还包括:
二值化阈值确定模块, 用于补偿后的灰度直方图或原始的灰度直方图中, 车牌字符的 灰度级小于车牌背景的灰度级, 从灰度直方图左侧向右侧开始累积灰度值对应的像素点数 目, 当累加值与总像素点数目比值为 0.3时, 将此时最后获取的一个像素点对应的灰度值作 为第一灰度值; 当累加值与总像素点数目比值为 0.4时, 将此时最后获取的一个像素点对应 的灰度值作为第二灰度值; 或者补偿后的灰度直方图或原始的灰度直方图中, 车牌字符的 灰度级大于车牌背景的灰度级, 从灰度直方图右侧向左侧开始累积灰度值对应的像素点数 目, 当累加值与总像素点数目比值为 0.3时, 将此时最后获取的一个像素点对应的灰度值作 为第一灰度值; 当累加值与总像素点数目比值为 0.4时, 将此时最后获取的一个像素点对应 的灰度值作为第二灰度值; 将位于所述第一灰度值和所述第二灰度值之间的灰度值, 按照 从小到大或从大到小的顺序进行排列, 并均分为至少六个数据区域; 将所述至少六个数据 区域中像素点最少的数据区域中的最小灰度值作为二值化阈值。
PCT/CN2011/081964 2011-11-01 2011-11-09 一种车牌图像定位的方法和装置 WO2013063820A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201110339877.XA CN103093181B (zh) 2011-11-01 2011-11-01 一种车牌图像定位的方法和装置
CN201110339877.X 2011-11-01

Publications (1)

Publication Number Publication Date
WO2013063820A1 true WO2013063820A1 (zh) 2013-05-10

Family

ID=48191243

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2011/081964 WO2013063820A1 (zh) 2011-11-01 2011-11-09 一种车牌图像定位的方法和装置

Country Status (2)

Country Link
CN (1) CN103093181B (zh)
WO (1) WO2013063820A1 (zh)

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108960244A (zh) * 2018-07-10 2018-12-07 北京智芯原动科技有限公司 一种车牌的字符投影分割方法及装置
CN109345462A (zh) * 2018-11-29 2019-02-15 北京京东尚科信息技术有限公司 图像处理方法、装置及计算机可读存储介质
CN109840699A (zh) * 2019-01-23 2019-06-04 深圳壹账通智能科技有限公司 合格投资者线上评估方法、装置、计算机设备和存储介质
CN110490207A (zh) * 2019-08-27 2019-11-22 河北科技大学 基于二值图像灰度共生矩阵的棒材端面字符图像识别方法
CN111898599A (zh) * 2020-07-10 2020-11-06 浙江大华技术股份有限公司 一种车牌图像处理方法及装置
CN112381088A (zh) * 2020-11-11 2021-02-19 上善智城(苏州)信息科技有限公司 一种油罐车车牌识别方法及系统
CN112435289A (zh) * 2020-12-14 2021-03-02 中国铁道科学研究院集团有限公司 钢轨接头轨缝宽度的检测方法及装置
CN112651985A (zh) * 2020-12-31 2021-04-13 康威通信技术股份有限公司 一种隧道巡检的里程标识牌定位方法及系统
CN113255657A (zh) * 2020-12-31 2021-08-13 深圳怡化电脑股份有限公司 票据表面刮痕检测方法、装置、电子设备、机器可读介质
CN113255641A (zh) * 2020-12-31 2021-08-13 深圳怡化电脑股份有限公司 一种图像识别的方法及装置、电子设备、存储介质
CN114998887A (zh) * 2022-08-08 2022-09-02 山东精惠计量检测有限公司 一种电能计量表智能识别方法
CN115100154A (zh) * 2022-06-29 2022-09-23 西安热工研究院有限公司 一种磁粉检测标准试片磁痕评估方法

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103488978B (zh) * 2013-09-26 2017-08-01 浙江工业大学 一种基于灰度跳变和字符投影间隔模式的车牌定位方法
CN106802144A (zh) * 2016-12-28 2017-06-06 深圳市美好幸福生活安全系统有限公司 一种基于单目视觉和车牌的车距测量方法
CN106846610B (zh) * 2017-01-20 2019-04-26 深圳怡化电脑股份有限公司 一种纸币版本识别方法及装置
CN108734161B (zh) * 2017-04-13 2022-01-25 深圳怡化电脑股份有限公司 冠字号区域的识别方法、装置、设备及存储介质
CN108038839B (zh) * 2017-11-23 2022-01-11 宁波大学 一种流水生产线上双绞线绞距实时检测方法
CN108305284B (zh) * 2018-02-28 2020-09-29 北京奇艺世纪科技有限公司 一种文字笔画宽度的确定方法及装置
CN108446697B (zh) * 2018-03-06 2019-11-12 平安科技(深圳)有限公司 图片处理方法、电子装置及存储介质
CN111199225B (zh) * 2018-11-20 2023-10-31 浙江宇视科技有限公司 车牌标定方法及装置
CN110414507B (zh) * 2019-07-11 2022-07-26 深圳智优停科技有限公司 车牌识别方法、装置、计算机设备和存储介质
CN112712091A (zh) * 2019-10-27 2021-04-27 北京易讯理想科技有限公司 图像特征区域检测方法
CN111126285A (zh) * 2019-12-22 2020-05-08 上海眼控科技股份有限公司 机动车牌证申请表的边界检测审核方法、设备及存储介质
CN113506283A (zh) * 2021-07-26 2021-10-15 浙江大华技术股份有限公司 图像处理方法及装置、存储介质、电子装置

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TW200529092A (en) * 2004-02-23 2005-09-01 Chunghwa Telecom Co Ltd Method for fetching plate area in vehicle image
KR20060094007A (ko) * 2005-02-23 2006-08-28 인천대학교 산학협력단 적응적 지역 이진화를 이용한 자동차 번호판 인식 시스템및 그 방법
CN1932838A (zh) * 2005-09-12 2007-03-21 电子科技大学 一种基于投影法和数学形态学的车牌提取方法
CN101739566A (zh) * 2009-12-04 2010-06-16 重庆大学 基于自适应投影模板法的车牌定位方法

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TW200802137A (en) * 2006-06-16 2008-01-01 Univ Nat Chiao Tung Serial-type license plate recognition system
CN101334837B (zh) * 2008-07-31 2012-02-29 重庆大学 一种多方法融合的车牌图像定位方法
CN102054169B (zh) * 2010-12-28 2013-01-16 青岛海信网络科技股份有限公司 车牌定位方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TW200529092A (en) * 2004-02-23 2005-09-01 Chunghwa Telecom Co Ltd Method for fetching plate area in vehicle image
KR20060094007A (ko) * 2005-02-23 2006-08-28 인천대학교 산학협력단 적응적 지역 이진화를 이용한 자동차 번호판 인식 시스템및 그 방법
CN1932838A (zh) * 2005-09-12 2007-03-21 电子科技大学 一种基于投影法和数学形态学的车牌提取方法
CN101739566A (zh) * 2009-12-04 2010-06-16 重庆大学 基于自适应投影模板法的车牌定位方法

Cited By (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108960244A (zh) * 2018-07-10 2018-12-07 北京智芯原动科技有限公司 一种车牌的字符投影分割方法及装置
CN109345462A (zh) * 2018-11-29 2019-02-15 北京京东尚科信息技术有限公司 图像处理方法、装置及计算机可读存储介质
CN109840699A (zh) * 2019-01-23 2019-06-04 深圳壹账通智能科技有限公司 合格投资者线上评估方法、装置、计算机设备和存储介质
CN110490207B (zh) * 2019-08-27 2023-07-18 河北科技大学 基于二值图像灰度共生矩阵的棒材端面字符图像识别方法
CN110490207A (zh) * 2019-08-27 2019-11-22 河北科技大学 基于二值图像灰度共生矩阵的棒材端面字符图像识别方法
CN111898599A (zh) * 2020-07-10 2020-11-06 浙江大华技术股份有限公司 一种车牌图像处理方法及装置
CN112381088A (zh) * 2020-11-11 2021-02-19 上善智城(苏州)信息科技有限公司 一种油罐车车牌识别方法及系统
CN112435289A (zh) * 2020-12-14 2021-03-02 中国铁道科学研究院集团有限公司 钢轨接头轨缝宽度的检测方法及装置
CN112435289B (zh) * 2020-12-14 2024-04-30 中国铁道科学研究院集团有限公司 钢轨接头轨缝宽度的检测方法及装置
CN113255657A (zh) * 2020-12-31 2021-08-13 深圳怡化电脑股份有限公司 票据表面刮痕检测方法、装置、电子设备、机器可读介质
CN113255641A (zh) * 2020-12-31 2021-08-13 深圳怡化电脑股份有限公司 一种图像识别的方法及装置、电子设备、存储介质
CN113255657B (zh) * 2020-12-31 2024-04-05 深圳怡化电脑股份有限公司 票据表面刮痕检测方法、装置、电子设备、机器可读介质
CN112651985A (zh) * 2020-12-31 2021-04-13 康威通信技术股份有限公司 一种隧道巡检的里程标识牌定位方法及系统
CN115100154A (zh) * 2022-06-29 2022-09-23 西安热工研究院有限公司 一种磁粉检测标准试片磁痕评估方法
CN115100154B (zh) * 2022-06-29 2024-02-27 西安热工研究院有限公司 一种磁粉检测标准试片磁痕评估方法
CN114998887A (zh) * 2022-08-08 2022-09-02 山东精惠计量检测有限公司 一种电能计量表智能识别方法
CN114998887B (zh) * 2022-08-08 2022-10-11 山东精惠计量检测有限公司 一种电能计量表智能识别方法

Also Published As

Publication number Publication date
CN103093181A (zh) 2013-05-08
CN103093181B (zh) 2016-04-27

Similar Documents

Publication Publication Date Title
WO2013063820A1 (zh) 一种车牌图像定位的方法和装置
US10896349B2 (en) Text detection method and apparatus, and storage medium
US8184859B2 (en) Road marking recognition apparatus and method
CN111382704B (zh) 基于深度学习的车辆压线违章判断方法、装置及存储介质
US9811921B2 (en) Apparatus and method for processing a depth image
CN109583280A (zh) 车道线识别方法、装置、设备及存储介质
WO2016065701A1 (zh) 图像文字识别方法及装置
CN106650739B (zh) 一种车牌字符切割新方法
CN104700062A (zh) 一种识别二维码的方法及设备
CN105303153A (zh) 一种车辆车牌识别方法及装置
CN110619333A (zh) 一种文本行分割方法、文本行分割装置及电子设备
CN112733703A (zh) 一种车辆停放状态检测方法及系统
CN110245544A (zh) 一种确定停车状态的方法及装置
CN110929544A (zh) 基于舌体轮廓线的舌体齿痕识别装置及方法
CN111598104A (zh) 一种车牌字符识别方法及其系统
CN111626225A (zh) 一种工位车辆的车牌识别方法、装置、设备及存储介质
CN112434583B (zh) 车道横向减速标线检测方法、系统、电子设备及存储介质
JP2015103249A (ja) 画像中のゼブラクロッシングの検出装置及び方法
CN110059695B (zh) 一种基于垂直投影的字符分割方法及终端
CN104408430B (zh) 一种车牌定位方法及装置
CN114926817B (zh) 识别车位的方法、装置、电子设备和计算机可读存储介质
CN108256475B (zh) 一种票据图像倒置检测方法
CN110363766A (zh) 马赛克检测方法、装置和电子设备
JP2004280713A (ja) 車番認識装置
CN111652143B (zh) 一种车辆检测方法、装置以及计算机存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 11874896

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 11874896

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 11874896

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 28/11/2014)

122 Ep: pct application non-entry in european phase

Ref document number: 11874896

Country of ref document: EP

Kind code of ref document: A1