KR101574874B1 - Method and apparatus for character segmentation of license plate - Google Patents

Abstract

The present invention relates to a method and apparatus for character division of a license plate,
The method comprising: determining a license plate type included in license plate images; A first processing step of recognizing and labeling each of the characters included in the number plate image according to a first adaptive heuristic division method if the number plate type is a black character type; And a second processing step of recognizing and labeling each of the characters included in the license plate image according to a second adaptive heuristic division technique if the license plate type is a white character type, So that it can perform an operation.

Description

TECHNICAL FIELD [0001] The present invention relates to a method and apparatus for dividing characters in a license plate,

The present invention relates to a vehicle license plate detection technology, and more particularly, to a method and apparatus for character registration of a license plate that can improve the detection performance of characters existing in a license plate image.

As the number of vehicles increases and the traffic environment becomes complicated, Intelligent Transport System (ITS) has been studied for efficient traffic management. It is composed of intelligent infrastructure system and intelligent vehicle system. plate recognition is a key part of intelligent infrastructure system and it is used in many applications such as highway toll payment system, parking lot access control, traffic volume monitoring and information management, and it plays an important role in performing simple repetitive tasks of people.

LPR is composed of three steps as shown in FIG. First, a vehicle license plate is detected in an image (S1). A second step is to divide the characters in the detected license plate. A third step is to recognize the divided characters.

Particularly, the field of interest in the present invention is a character segmentation step, which is an important step in which characters in a license plate are essentially divided for character recognition. Detected license plates are subject to many difficulties due to non-uniform pixel distribution, illumination changes, and noise. In order to solve this problem, a binary image is constructed based on an adaptive binarization method which is processed locally, so that a split study is being conducted.

First, the first method is a character segmentation method using pixel connectivity. It is able to know the position, size, and aspect ratio by labeling through the information of pixels connected in binary. It is simple to implement and has a strong vertex in the rotated license plate, but it has a drawback in that it can fail to divide if the pixels per character are attached or not connected well.

The second method is a character segmentation method using projection, which is a classical method of character segmentation and is now widely used. Projection in the vertical direction and horizontal direction after preprocessing through noise elimination and morphology operation in the binoculars of the license plate has the advantage of dividing the characters independently, but if the license plate is rotated or the noise is heavy, the performance may be greatly affected .

As another method, there is a character division method using prior knowledge. In the license plate, the position, size, and the distance between the font and the character are determined. However, it is limited and susceptible to change, which can lead to incorrect division results.

On the other hand, Korean license plates can be broadly divided into black and white character types. In case of plates with white background and yellow background, when converted to gray image, characters have relatively low values and green plates have high values. Can be limited. 2 shows the type of the Korean license plate described above. However, the conventional character segmentation method has a problem in that it can not support all of these various license plate types.

In order to solve the above problems, the present invention provides a method and apparatus for dividing characters of a license plate, which can always perform a reliable character division operation irrespective of a license plate type.

The objects of the present invention are not limited to the above-mentioned objects, and other objects not mentioned can be clearly understood by those skilled in the art from the following description.

As a means for solving the above-mentioned problems, according to an embodiment of the present invention, there is provided a method for determining a license plate type, comprising: determining a license plate type included in license plate images; A first processing step of recognizing and labeling each of the characters included in the number plate image according to a first adaptive heuristic division method if the number plate type is a black character type; And a second processing step of recognizing and labeling each of the characters included in the license plate image according to a second adaptive heuristic division technique if the license plate type is a white character type.

Wherein the determining step includes performing adaptive histogram smoothing on the license plate image and removing noise included in the adaptive histogram smoothed license plate image; Setting a center region of the noiseless license plate image as a region of interest; Binarizing the region of interest according to a global thresholding technique to classify all pixels located within the region of interest into a white pixel and a black pixel; And determining that the license plate type is a black character type if the number of white pixels is greater than a black pixel, and otherwise determining that the license plate type is a white character type.

의 수학식을 사용하는 바이레터럴 필터를 통해 잡음 제거 동작을 수행하며, 상기

는 필터링된 번호판 영상, 상기

는 필터링전 번호판 영상, 상기

는 픽셀의 좌표, 상기

는 윈도우, 상기

은 픽셀 값의 차이를 스무딩하기 위한 범위 커널, 상기

는 좌표 차이를 스무딩하기 위한 공간 커널인 것을 특징으로 한다. The step of removing the noise included in the license plate image

A noise removal operation is performed using a bilateral filter using the equation

The filtered license plate image,

A pre-filtering license plate image,

The coordinates of the pixel,

Window,

A range kernel for smoothing the difference in pixel values,

Is a spatial kernel for smoothing the coordinate difference.

And the global threshold value processing method is implemented by an image quality binarization method.

Wherein the first processing step comprises: binarizing the license plate image according to a local threshold value processing technique and then performing blob filtering; Removing the lower outer edge region and the upper outer edge region of the blob-filtered number plate image; And recognizing and labeling each character using the RL (Right Left) technique and the LR (Left Right) technique.

,

,

,

의 수식으로 구현되는 적응 문턱치 처리 기법으로 구현되며, 상기

은 번호판 영상,

는 픽셀의 위치,

은 문턱치,

은 이진화된 번호판 영상,

는 원도우내 평균,

는 원도우내 표준편차, 상기 k와 상기 R는 계산 계수인 것을 특징으로 한다. The local threshold processing technique

,

,

,

Is implemented with an adaptive threshold value processing technique implemented as an equation of

The license plate image,

The position of the pixel,

The threshold,

A binned license plate image,

The average in the window,

Is a standard deviation in a window, and k and R are calculation coefficients.

는 원도우내 평균,

는 원도우내 표준편차 계산 시에, 적분 영상을 사용하는 것을 특징으로 한다. The local thresholding technique may include

The average in the window,

Is characterized by using an integral image at the time of calculating standard deviation within a window.

The step of recognizing and labeling the character may further include the step of analyzing the labeling result based on the predetermined inter-character distance and the position reference value to detect and remove the erroneously labeled blobs.

The step of recognizing and labeling the character may further include labeling a portion between the second and third digits by locally morphing the digits into a single lump.

Separating the lower line image and the upper line image of the license plate image; Performing a blob filtering on the bottom line image, recognizing and labeling the numbers included in the bottom line image through an RL (Right Left) technique, and outputting the number; Determining a region of interest of the upper line image based on a position value of the numbers included in the lower line image; And a step of morphology-calculating the ROI, recognizing and labeling the numbers and characters included in the ROI, and outputting the ROI.

In the step of separating the lower line image and the upper line image, a bottom line image is obtained through a BU (bottom up) technique, and an upper line image is obtained through a horizontal projection technique.

According to another aspect of the present invention, there is provided a means for solving the above problems, comprising: a license plate type determining unit for identifying a license plate type included in license plate images; A first character division unit for recognizing and labeling each of the characters included in the number plate image according to a first adaptive heuristic division technique if the number plate type is a black character type; And a second character division unit for recognizing and labeling each of the characters included in the license plate image according to a second adaptive heuristic division technique if the license plate type is a white character type.

According to the present invention, different character segmentation methods are performed according to the license plate type, thereby ensuring more improved character segmentation performance. In addition, the overall performance of license plate recognition (LPR) can be improved by securing character segmentation performance.

1 is a view for explaining a general license plate recognition method.
2 is a view for explaining the license plate type of the Republic of Korea.
FIG. 3 is a diagram illustrating a method of dividing characters in a license plate according to an embodiment of the present invention.
4 is a diagram for explaining an adaptive histogram smoothing method according to an embodiment of the present invention.
FIG. 5 is a diagram for explaining an adaptive bi-linear filtering method according to an embodiment of the present invention. Referring to FIG.
FIG. 6 is a diagram for explaining the black character segmentation step according to the first adaptive heuristic segmentation algorithm according to an embodiment of the present invention in more detail.
7 is a view for explaining an integral image for a character division method of a license plate according to an embodiment of the present invention.
FIG. 8 is a diagram illustrating a result of comparing Otsu threshold processing, which is a global threshold processing technique according to an embodiment of the present invention, and an adaptive threshold treated plate, which is a local threshold processing technique.
FIG. 9 is a diagram illustrating an effect of using an integral image to determine an average and a standard deviation in an adaptive threshold processing according to an exemplary embodiment of the present invention. Referring to FIG.
10 is a diagram for explaining a method of using the BU algorithm according to an embodiment of the present invention.
11 is a view for explaining a method of using a horizontal projection algorithm according to an embodiment of the present invention.
12 is a diagram for explaining a method of using the RL algorithm according to an embodiment of the present invention.
13 is a diagram for explaining a method of using the LR algorithm according to an embodiment of the present invention.
FIG. 14 is a diagram for explaining a method of processing a local morphology according to an embodiment of the present invention.
15 is a diagram for explaining a white character segmentation step according to a second adaptive heuristic segmentation algorithm according to an embodiment of the present invention in more detail.
16 is a diagram showing examples of character division result according to an embodiment of the present invention.
17 is a diagram showing a character division device of a license plate according to an embodiment of the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS The above and other objects, features and advantages of the present invention will be more apparent from the following detailed description taken in conjunction with the accompanying drawings, in which: FIG. In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear.

The following terms are defined in consideration of the functions of the present invention, and these may be changed according to the intention of the user, the operator, or the like.

The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. These embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art to which the present invention pertains. Only. Therefore, the definition should be based on the contents throughout this specification.

FIG. 3 is a diagram illustrating a method of dividing characters in a license plate according to an embodiment of the present invention.

Referring to FIG. 3, the character segmentation method of the present invention includes a license plate type determination step S10, a step S20 of dividing characters included in a black character type license plate according to a first adaptive heuristic segmentation technique, A step S30 of dividing the characters included in the white letter type number plate according to the division technique, and a step S40 of outputting the character division result.

That is, considering that the Korean license plate is divided into two types, the present invention proposes optimal character division methods for each of the two types, thereby ensuring optimal detection performance for both types of Korean license plates .

Hereinafter, the character division method of the license plate will be described in more detail with reference to FIG. 4 to FIG.

4 is a view for explaining the license plate type determination step according to an embodiment of the present invention in more detail.

For reference, the size of an input image (that is, a license plate image) may differ from image to image, and in this case, the image processing load is unnecessarily increased.

In the present invention, the size of the license plate image is normalized to a preset value through step S11. The image size normalization may be performed by a method such as cutting the image area or changing the resolution of the image.

In step S12, adaptive histogram smoothing is performed on the license plate image. In the case of the license plate image provided for the plate type determination step, there may be a case where the distribution of the pixels is uneven as shown in FIG. 5 (a). However, since the adaptive histogram smoothing method is processed locally rather than globally, the smoothing processing can be effectively performed even if the distribution of pixel values is shifted. Thus, in the present invention, the pixel distribution of the license plate image is uniformized through the adaptive histogram smoothing method as shown in FIG. 5 (b).

In step S13, since adaptive histogram smoothing is applied, the noise included in the license plate image is removed by bilaterally filtering the license plate image in consideration of the fact that a slight noise occurs because the contrast is increased. A by-letter filter is a non-linear method that preserves edge portions while reducing noise and smoothing. This filter is constructed as shown in the following Equation (1).

는 필터링된 번호판 영상,

는 필터링전 번호판 영상,

는 픽셀의 좌표,

는 윈도우,

은 픽셀 값의 차이를 스무딩하기 위한 범위 커널,

는 좌표 차이를 스무딩하기 위한 공간 커널을 각각 의미한다. At this time,

A filtered license plate image,

The pre-filtering plate image,

The coordinates of the pixel,

Windows,

Is the range kernel for smoothing differences in pixel values,

Respectively denote a spatial kernel for smoothing the coordinate difference.

5 (a) shows the pre-applied image of the bi-literal filter, and FIG. 5 (b) shows the image after applying the bi-linear filter. By referring to this, it is possible to obtain a clearer and clean image .

In step S14, the center plate of the license plate-filtered license plate image is acquired as a region of interest, considering that the license plate is located at the middle portion of the license plate image and the background region is wider than that of the character portion.

In step S15, the region of interest (Otsu) is image-binarized to divide all the pixels located in the region of interest into white pixels and black pixels.

In the case of white or yellow plates, pixels located in the background region have a value of 255 (that is, classified as white pixels), pixels located in the character region have a value of 0 (that is, classified as black pixels) The number of the white pixels and the number of the black pixels can be compared to determine the type of the license plate.

That is, in step S16, the number of white pixels is compared with the number of black pixels. If the number of white pixels is larger than the number of black pixels, it is determined that the license plate image is a black character type. Type.

6 is a diagram for explaining the black character division step according to the first adaptive heuristic division method according to an embodiment of the present invention in more detail.

First, in step S21, image binarization is performed through an adaptive threshold value processing method.

은 번호판 영상,

는 픽셀의 위치,

은 문턱치,

은 이진화된 번호판 영상을 의미한다. 지역 적응 문턱치 처리 방법은 각 픽셀에서 문턱치를 계산하여 0 혹은 255 값을 갖는 이진영상을 만든다. The adaptive threshold processing method is robust to illumination because it processes the threshold value locally. This can be expressed as in Equation 2,

The license plate image,

The position of the pixel,

The threshold,

Means a binned license plate image. The local adaptive threshold processing method generates a binary image with 0 or 255 values by calculating the threshold value at each pixel.

은 윈도우내의 평균

, 표준편차

, 계산 계수 k, R를 이용하여 문턱치를 구하게 된다. Equation (3) below is a formula for calculating the threshold value,

The average within the window

, Standard Deviation

, The calculation coefficients k and R are used to obtain the threshold value.

Equations (4) and (5) below are mathematical expressions for obtaining the average and variance of the original image. In this process, the mean and standard deviation can be obtained at a high speed using the integral image of Equation (6) and (7) below. In this case, w is a window size in the original image.

FIG. 8 is a graph comparing the Otsu threshold value processing, which is a global threshold value processing method, and the adaptive threshold value processing, which is a local threshold value processing method, and it can be seen that the adaptive threshold value processing is robust.

FIG. 9 shows the effect of using the integral image in determining the mean and standard deviation during adaptive threshold processing. In FIG. 9, the solid line represents the case where the integral image is used and the dotted line represents the case where the integral image is not used. When the integral image is not used, the calculation amount (i.e., calculation time) increases as the size of the local window increases .

When the generation of the binary image is completed, the binary image is filtered using the reference information about the size and aspect ratio of the labeled blobs in step S22. If the remaining blobs are a predetermined number (for example, seven or less), it is determined that the image is a white character type, not a black character type, and the second adaptation And enters the white character segmentation step S30 according to the heuristic division technique.

In step S23, an unnecessary part existing at the lower end of the license plate image is removed through a BU (bottom up) technique as shown in FIG. BU is a technique that removes labeled bobbins from bottom to top until they are in a straight line. This eliminates unnecessary parts of the license plate.

In step S24, the upper edge image is removed through a horizontal projection technique, and noise remaining in the license plate image is removed through adaptive binarization.

In the license plate image, there is a portion with margins between the characters and the frame. Thus, in the present invention, the remaining pixels are projected in the horizontal direction and are searched in the downward direction as shown in FIG. 11 (a). Here, using the average of the pixel distributions, a point having the greatest gradient is found, and the upper image is removed based on this point, thereby obtaining only a portion having a character as shown in FIG. 11 (b). In addition, adaptive binarization is performed to remove noise remaining in the image as shown in FIG. 11 (c).

In step S25, blob filtering is performed again to remove noise.

In step S26, each character is recognized and labeled by using an RL (Right Left) technique and an LR (Left Right) technique based on preset character type information. Then, based on the preset distance between characters and the reference value of the position, analyzing the distance and positional relationship between the labeled characters detects and removes the mislabeled blobs.

12, there are two numbers on the left and one letter and four numbers on the right, and they are separated from each other. In the case of the RL method, look at the distance relationship of the five blobs from right to left, move to the left until the condition is met, and check to determine the four numbers on the right.

The LR method uses prior knowledge as well as the LR method. The search range is determined by using the distance between the first and fourth characters of the four characters obtained by applying the LR method in the previous step, then moved from left to right as shown in FIG. 13, and the blob relation is checked, Two characters and one character are determined. Here, if the condition is not satisfied, it is regarded as a two-line license plate (for example, a green license plate) and a white character division technique is applied.

Finally, in step S26, the character region portions existing between the second and third digits are locally morphologically processed into a single lump to be labeled. For example, the words "ga, ga, da, la" and so on are separated, although they are located between the second and third digits of the license plate. In the present invention, in order to prevent the occurrence of the problem in advance, the part between the second and third numbers is morphologically subjected to morphology processing as shown in FIG. 14, and the labeling operation for the characters existing in the area is also stably performed .

15 is a diagram for explaining a white character segmentation step according to a second adaptive heuristic segmentation technique according to an embodiment of the present invention in more detail.

The white letter type plate consists of two lines, as described above. Accordingly, in the present invention, after dividing the upper row and the lower row, a character division operation is performed for each of the upper row and the lower row.

First, in step S31, image binarization is performed through the above-described adaptive threshold value processing method.

In step S32, the bottom line image of the license plate image is acquired through the BU technique. In step S33, the upper line image of the license plate image is acquired through the horizontal projection technique.

In step S34, the bottom line image is blob filtered to remove noise. In step S35, four numbers are recognized and labeled using the RL technique based on the prior knowledge.

In step S35, the number of the upper line image and the area of interest in which the character is located are identified based on the positions of the recognized numbers in step S34. In step S36, the upper edge image is removed by applying a horizontal projection again to the area of interest , Reducing the image size.

In step S37, the positions of the numbers and the characters are obtained by labeling through the morphology operation.

FIG. 16 is a diagram illustrating examples of character division according to an exemplary embodiment of the present invention. Referring to FIG. 16, all characters included in a license plate image can be recognized regardless of the license plate type.

17 is a diagram showing a character division device of a license plate according to an embodiment of the present invention.

17, the character segmentation apparatus of the present invention includes a license plate type determination unit 10 for identifying a license plate type included in a license plate image, and a license plate type determination unit 10 for determining whether the license plate type is a black character type according to a first adaptive heuristic division technique A first character division unit 20 for recognizing and labeling each character included in the license plate image, and a second character division unit 20 for recognizing each character included in the license plate image according to a second adaptive heuristic division technique, And a second character division unit 30 for labeling and the like.

In other words, the apparatus of the present invention proposes two types of adaptive heuristic division techniques considering that Korean license plates are classified into two types, and through these, Thereby performing a character division operation.

The character dividing device of the present invention configured as described above may be configured independently as one hardware device, but may be implemented as an embedded device embedded in an LPR device that performs both a license plate detecting function, a license plate character dividing function, and a divided character recognizing function It may be implemented.

The computer-readable recording medium on which the program commands are recorded may be a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, Media storage devices.

The computer-readable recording medium on which the above-described program is recorded may be distributed to a computer apparatus connected via a network so that computer-readable codes can be stored and executed in a distributed manner. In this case, one or more of the plurality of distributed computers may execute some of the functions presented above and send the results of the execution to one or more of the other distributed computers, The computer may also perform some of the functions described above and provide the results to other distributed computers as well.

A computer capable of reading a recording medium on which a method of dividing a license plate of a vehicle according to each embodiment of the present invention and an application that is a program for driving the apparatus are recorded can be used not only in a general PC such as a general desktop or a notebook computer, A tablet PC, a PDA (Personal Digital Assistants), and a mobile terminal, and in addition, it should be interpreted as all devices capable of computing.

While the present invention has been described in connection with what is presently considered to be the most practical and preferred embodiments, it is to be understood that the invention is not limited to the disclosed embodiments. That is, within the scope of the present invention, all of the components may be selectively coupled to one or more of them. In addition, although all of the components may be implemented as one independent hardware, some or all of the components may be selectively combined to perform a part or all of the functions in one or a plurality of hardware. As shown in FIG. The codes and code segments constituting the computer program may be easily deduced by those skilled in the art. Such a computer program can be stored in a computer-readable storage medium, readable and executed by a computer, thereby realizing an embodiment of the present invention. As a storage medium of the computer program, a magnetic recording medium, an optical recording medium, or the like can be included.

It is also to be understood that the terms such as " comprises, "" comprising," or "having ", as used herein, mean that a component can be implanted unless specifically stated to the contrary. But should be construed as including other elements. All terms, including technical and scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs, unless otherwise defined. Commonly used terms, such as predefined terms, should be interpreted to be consistent with the contextual meanings of the related art, and are not to be construed as ideal or overly formal, unless expressly defined to the contrary.

The foregoing description is merely illustrative of the technical idea of the present invention, and various changes and modifications may be made by those skilled in the art without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are intended to illustrate rather than limit the scope of the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be construed according to the following claims, and all technical ideas within the scope of equivalents should be construed as falling within the scope of the present invention.

Claims (13)

A determining step of determining a license plate type included in the license plate image;
A first processing step of recognizing and labeling each of the characters included in the number plate image according to a first adaptive heuristic division method if the number plate type is a black character type; And
And a second processing step of recognizing and labeling each of the characters included in the license plate image according to a second adaptive heuristic division technique if the license plate type is a white character type,
The second processing step
Separating a lower line image and an upper line image of the license plate image;
Performing a blob filtering on the bottom line image, recognizing and labeling the numbers included in the bottom line image through an RL (Right Left) technique, and outputting the number;
Determining a region of interest of the upper line image based on a position value of the numbers included in the lower line image; And
And performing a morphological operation on the ROI to recognize and label and output the numbers and characters included in the ROI. 2. The method of claim 1,
Performing adaptive histogram smoothing on the license plate image, and removing noise included in the adaptive histogram smoothed license plate image;
Setting a center region of the noiseless license plate image as a region of interest;
Binarizing the region of interest according to a global thresholding technique to classify all pixels located within the region of interest into a white pixel and a black pixel; And
Determining that the license plate type is a black character type if the number of white pixels is greater than a black pixel, and otherwise determining that the license plate type is a white character type. 3. The method of claim 2, wherein removing the noise included in the license plate image

Lt; RTI ID = 0.0 > a < / RTI > by-
remind

The filtered license plate image,

A pre-filtering license plate image,

The coordinates of the pixel,

Window,

A range kernel for smoothing the difference in pixel values,

Is a space kernel for smoothing a coordinate difference. 4. The method of claim 3, wherein the global thresholding technique
And the image is binarized by the image of the person. 2. The method of claim 1, wherein the first processing step
Binarizing the license plate image according to a local threshold value processing method, and then performing blob filtering;
Removing the lower outer edge region and the upper outer edge region of the blob-filtered number plate image; And
And recognizing and labeling each character using an RL (Right Left) technique and an LR (Left Right) technique. 6. The method of claim 5,

And is implemented as an adaptive threshold value processing technique,
remind

The license plate image,

The position of the pixel,

The threshold,

A binned license plate image,

The average of the original images in the window,

Is a standard deviation in a window of the original image, w is a window size in the original image, and k and R are calculation coefficients. 7. The method of claim 6, wherein the local thresholding technique
remind

The average in the window,

Wherein an integral image is used in calculating a standard deviation of a window. 6. The method of claim 5, wherein recognizing and labeling the character
Further comprising the step of analyzing the labeling result based on the preset inter-character distance and position reference value to detect and remove the mislabeled blobs. 6. The method of claim 5, wherein recognizing and labeling the character
Further comprising the step of locally performing morphological processing of the character area portion existing between the second and third digits so as to form a single lump and labeling. delete The method of claim 1, wherein separating the lower line image and the upper line image comprises:
A bottom line image is obtained through a BU (bottom up) technique, and an upper line image is obtained through a horizontal projection technique. A computer-readable recording medium having recorded thereon a program for executing the character division method of the license plate according to any one of claims 1 to 9 and 11. A license plate type determination unit for identifying license plate types included in license plate images;
A first character division unit for recognizing and labeling each of the characters included in the number plate image according to a first adaptive heuristic division technique if the number plate type is a black character type; And
And a second character division unit for recognizing and labeling each of the characters included in the license plate image according to a second adaptive heuristic division technique if the license plate type is a white character type,
The second character dividing unit
The bottom line image of the license plate image is separated from the upper line image, and the bottom line image is subjected to blob filtering, and the numbers included in the bottom line image are recognized and labeled through the RL (Right Left) Recognizes the region of interest of the upper line image based on the position value of the numbers included in the line image, and performs morphology calculation of the region of interest, recognizes and labels the numbers and characters included in the region of interest, and outputs the number and characters. Character splitting device of license plates.

Images