KR101218302B1 - Method for location estimation of vehicle number plate - Google Patents

Abstract

An object of the present invention is to provide a vehicle license plate position estimation method and a vehicle license plate position estimation device capable of accurately estimating the position of a license plate of a vehicle in an image acquired from an imaging device. In the vehicle license plate position estimation method of a parking lot monitoring system provided with an imaging device, (a) by analyzing the image picked up from the image pickup device, a candidate portion in which the vehicle license plate is likely to be located among the portions of the picked-up image. And (b) processing each video signal of the selected candidate parts into a frequency use classifier and a spatial information use classifier, respectively; and (c) using the frequency use classifier and the spatial information. Normalizing and weighting the similarity values calculated by the classifier for mutual comparison; and (d) Determining a candidate portion having the largest value among the normalized and weighted similarity values as a license plate position by collecting the normalized and weighted similarity values and comparing them with each other; and A vehicle license plate position estimation apparatus for performing the vehicle license plate position estimation method is provided.

Description

Method for location estimation of vehicle number plate}

1 is a diagram illustrating a configuration of an illegal parking control system according to an embodiment of the present invention.

2 is a block diagram showing the configuration of a center apparatus according to an embodiment of the present invention.

3 is a flowchart illustrating a process of performing a vehicle license plate position estimation method according to an embodiment of the present invention.

4 is a diagram illustrating a state in which candidate parts having a high possibility of being located in a vehicle license plate are selected from an image photographed by an image capturing apparatus according to an exemplary embodiment of the present invention.

Description of the Related Art [0002]

100: illegal parking control system 110-1, ..., 110-n: camera

120: network 130: center gear

131: user interface unit 132: surveillance area setting unit

133: vehicle recognition unit 134: license plate recognition unit

135: camera control unit 136: enforcement unit

137: storage unit 138: network connection

140: vehicle region 141, 142, 143, 144: candidate portion

The present invention relates to a method and an apparatus for estimating the position of a license plate which is a step before the step of recognizing a character of the license plate.

Recently, as the number of vehicles has increased due to industrial development and income increase, research on illegal parking control systems has been vigorous.

The conventional illegal parking control system includes an imaging device, a control system, and the like. The operation process is as follows.

When the vehicle stops in the monitoring area where the illegal parking control system is installed in advance, the installed imaging device captures the parked vehicle and acquires an image of the vehicle. Then, the crackdown system estimates the position of the license plate of the vehicle in the acquired image and recognizes the letters on the license plate to perform crackdown.

Here, as a conventional method of estimating the position of the license plate of the vehicle in the acquired image, i) a method using the texture of the vehicle license plate, ii) a method using color information and the like have been used.

However, the above-mentioned conventional methods are not license plates, but it is difficult to accurately estimate the position of the license plate of the vehicle when a portion having a shape similar to the license plate is photographed, or when the color of the license plate is changed or the external lighting conditions are changed. There were difficult problems.

Therefore, there is a need for developing a new technology for more accurately estimating the position of the license plate of the vehicle in the image acquired from the imaging device.

It is a main object of the present invention to provide a vehicle license plate position estimation method and a vehicle license plate position estimation device capable of accurately estimating the position of a license plate of a vehicle in an image acquired from an imaging device.

According to the present invention, there is provided a method for estimating a vehicle license plate position of a parking lot monitoring system having an imaging device, comprising: (a) analyzing an image picked up from the imaging device, whereby a vehicle license plate is likely to be located in a portion of the image picked up. Selecting a plurality of candidate portions; and (b) processing each video signal of the selected candidate portions with a frequency use classifier and a spatial information use classifier, respectively; and (c) the frequency use classifier and the space. Normalizing and weighting the similarity values calculated by the information use classifier for mutual comparison; and (d) collecting the normalized and weighted similarity values and comparing them with each other to thereby compare the normalized and weighted similarity values. Determining the candidate portion having the largest value among the values as the license plate position; A position estimation method is disclosed.

Here, in order to select the candidate parts in step (a), an image filtering method using a color coordinate system may be applied to the captured image.

Here, in order to select the candidate parts in step (a), an image filtering method using the texture of the captured image may be applied.

Here, the image filtering methods may use a blob labeling method of ignoring blobs of a predetermined size or less than a predetermined size.

Here, between the steps (a) and (b), (e) normalizing the size of the image of the selected candidate parts; and (f) improving the image quality of the images passed through the step (e). Steps may further include.

Here, in the step (f), the image quality improvement may use a contrast enhancement method.

Here, in the step (f), the image quality improvement may use a brightness compensation method.

Here, in the step (b), the process of processing the video signal of each of the selected candidate parts with the frequency utilization classifier includes the frequency values of the video signal of the reference license plate input in advance and the video signals of the selected candidate parts. Comparing the frequency values of may include calculating a similarity value.

Here, in the step (b), the process of processing the image signal of each of the selected candidate parts with the spatial information use classifier may include gradation values of pixels of an image of a pre-input reference number plate and images of the selected candidate parts. Comparing the grayscale values of the pixels of, may include calculating a similarity value.

In addition, the present invention discloses a vehicle license plate position estimation apparatus for performing the vehicle license plate position estimation method.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram illustrating a configuration of an illegal parking control system according to an embodiment of the present invention.

Referring to FIG. 1, the illegal parking control system 100 according to the present invention includes a plurality of cameras 110-1 to 110-n, a communication network 120, and a center device 130.

The plurality of cameras 110-1 to 110-n are installed in the intermittent subject area as an imaging device, and two or more cameras may be installed in the same subject area.

Each camera 110-1 to 110-n is assigned its own surveillance area. Each camera 110-1 to 110-n may employ any camera having a function of transmitting the captured image to the center device 130, but the pan / tilt / zoom can be freely taken. It is preferable to employ a web camera for digitizing and transmitting. At this time, each camera (110-1 to 110-n) is controlled by a control command received from the center device 130, the monitoring area of each camera (110-1 to 110-n) is the center device ( 130).

The center device 130 has a function of controlling each of the cameras 110-1 to 110-n connected through the communication network 120, which can be a wired or wireless Internet, a mobile communication network, or the like. 1 to 110-n), after recognizing the illegal parking vehicle on the basis of the image received from, performs a function of generating a fine bill.

2 is a block diagram showing the detailed configuration of the center apparatus 130. As shown in FIG.

Referring to FIG. 2, the center device 130 includes a user interface unit 131, a surveillance area setting unit 132, a vehicle recognition unit 133, a license plate recognition unit 134, a camera control unit 135, and a enforcement unit. 136, a storage unit 137, and a network connection unit 138.

The user interface unit 131 outputs images captured by the cameras 110-1 to 110-n and state information of the center apparatus 130, and is a means for receiving a command from an operator.

The surveillance region setting unit 132 sets the surveillance region of each of the cameras 110-1 to 110-n that are connected. The surveillance area may be configured as a wide area surveillance area of the front lower surveillance area, the lower lower surveillance area, and the rear lower surveillance area based on the point where the camera is installed.

The vehicle recognizing unit 133 extracts a vehicle area through image processing on an image received in real time from each of the cameras 110-1 to 110-n installed at an intermittent place through the network connection unit 138. That is, the vehicle recognition unit 133 estimates the vehicle area from the received image based on the symmetry of the headlights, the presence of shadows, the presence of a window shape, and the like.

The vehicle area estimated by the vehicle recognition unit 133 is displayed as the upper left and lower right coordinates (or the upper right and lower left coordinates) of the area recognized as the vehicle in the received image. At this time, the coordinates representing the vehicle area are calculated based on the resolution of the received image or calculated by matching the received image on a predetermined coordinate. The vehicle recognizing unit 133 generates vehicle area recognition information including an extraction time of the vehicle area, coordinates of the extracted vehicle area, and lower monitoring area identification information for each lower monitoring area, and stores the vehicle area recognition information in the storage unit 137. .

The license plate recognition unit 134 estimates the position of the license plate from the extracted vehicle region or the received image, and determines the vehicle number in the license plate region where the position is estimated. That is, the license plate recognition unit 134 estimates the position of the license plate of the vehicle in the photographed image, and then performs a function of recognizing a character in an area of the license plate whose position is estimated. Here, a method of estimating the position of the license plate of the vehicle by the license plate recognition unit 134 will be described later.

The center device 130 according to the present embodiment includes both the vehicle recognition unit 133 and the license plate recognition unit 134, but the present invention is not limited thereto. That is, according to the present invention, the center apparatus may include only the license plate recognition unit without the vehicle recognition unit. In that case, the license plate recognition unit searches and recognizes the position of the license plate of the vehicle directly from the captured image.

The camera controller 135 controls the operations of the cameras 110-1 to 110-n that are connected. The operator located in the center may adjust the pan / tilt / zoom of each of the cameras 110-1 to 110-n by using the user interface unit 131, and the camera controller 135 may control the control command received from the operator. Camera control information including the horizontal angle, the vertical angle, the zoom level, and the monitoring time of the corresponding camera is generated and transmitted to the control target camera through the network connection unit 138. The camera to be controlled operates based on the camera control information received from the center apparatus 130. When the camera control information is setting information for the surveillance area, the camera to be controlled stores the received camera control information in a storage unit provided therein.

The enforcement enforcement unit 136 processes the enforcement service for the illegal parking vehicle. The enforcement service by the enforcement enforcement unit 136 may be performed as follows.

That is, if the vehicle area is recognized at the same position for a predetermined time (for example, 5 seconds) or more based on the vehicle area recognition information generated by the vehicle recognition unit 133, the enforcement unit 136 may not be assigned to the vehicle area. The existing vehicle is regarded as a parked vehicle and the corresponding vehicle area recognition information is recorded in the parked vehicle database.

When the vehicle is stopped, the enforcement unit 136 instructs the license plate recognizing unit 134 to estimate the license plate position from the vehicle area to recognize the letters on the license plate, and generates the license plate area generated by the license plate recognition unit 134. Record the recognition information in the license plate database.

The enforcement enforcement unit 136 responds when the illegal parking regulation reference time (for example, 5 minutes) has elapsed from the discovery time of the vehicle being parked (that is, when the license plate area recognition information of the vehicle is recorded in the license plate database). The camera to capture the corresponding lower surveillance area. Such camera control may be configured to be performed by the camera controller 135 by driving the intermittent execution unit 136.

When photographing of the lower surveillance area is performed, the vehicle recognition unit 133 extracts the vehicle area from the image received from the camera. The enforcement unit 136 determines that the vehicle is an illegal parking vehicle if the extracted vehicle region matches the vehicle region included in the vehicle region recognition information recorded in the parking vehicle database.

Next, the enforcement unit 136 extracts the license plate area from the vehicle area extracted by the license plate recognition unit 134, estimates the vehicle number of the vehicle, and matches the vehicle number of the vehicle stored in the license plate database unit. Control to check whether or not. In this case, camera control information corresponding to the coordinates of the vehicle area and the license plate area extracted by the vehicle recognition unit 133 and the license plate recognition unit 134 may be used for controlling the camera. If it is confirmed that the same vehicle, the enforcement unit 136 may control the control point, the vehicle number, the storage location of the captured image at the start of illegal parking control, the storage location of the captured image at the elapsed time of the illegal parking control reference time, and the extracted license plate. An intermittent vehicle information including the coordinates of the area and the lower surveillance area identification information is generated and recorded in the intermittent vehicle database.

In addition, the enforcement unit 136 creates a charging document for the vehicle identified as being illegally parked based on the enforcement vehicle information. At this time, the billing document describes the vehicle number, the crackdown time of the vehicle (start time and stop time of the vehicle), the image at the time of crackdown (the start point of the parking lot and the exceeding point of the illegal parking control criterion), and the place.

The storage unit 137 includes camera control information, vehicle area recognition information, license plate area recognition information, and parking vehicle data for capturing a lower surveillance area constituting a surveillance area corresponding to each of the cameras 110-1 to 110-n. The base, the vehicle number database, the intermittent vehicle database, and the like are stored.

The network connection unit 138 is connected to a wired or wireless communication network to provide a means for transmitting and receiving data with each of the cameras 110-1 to 110-n. The center device 120 transmits camera control information to each of the cameras 110-1 through 110-n through the network connection unit 138, and displays images captured from each of the cameras 110-1 through 110-n. Receive.

Examples of network protocols supported by the network connection unit 138 include TCP / IP, HTTP, ARP, PARP, ICMP, DHCP, FTP, SMTP, and the like, and communication lines include xDSL, Ethernet, and LAN. The cameras 110-1 to 110-n and the center device 130 may communicate by various communication methods. For example, the xDSL modem may include the cameras 110-1 to 110-n and the center device 130. It can be employed as a communication means between.

Hereinafter, a method of estimating the position of the license plate of the vehicle by the license plate recognition unit 134 will be described in detail with reference to FIGS. 3 and 4.

After the vehicle region is extracted by the vehicle recognition unit 133 through image processing on the image received in real time from each of the cameras 110-1 to 110-n, the position of the license plate can be estimated from the extracted vehicle region. do.

In the present embodiment, once the vehicle area is extracted, the license plate position is searched for, but the present invention is not limited thereto. That is, as described above, the present invention may not include the vehicle recognition unit 133, in which case, the position of the license plate of the vehicle is directly searched from the captured image.

In detail, first, the license plate recognition unit 134 analyzes the images captured by the cameras 110-1 through 110-n to select a plurality of candidate portions having a high possibility of the vehicle license plate being located among the captured images. (Step S1).

In this case, an image filtering method using a color coordinate system and an image filtering method using a texture may be used as an image analysis method for selecting a plurality of candidate parts.

An image filtering method using a color coordinate system may use the method disclosed in Japanese Patent Laid-Open No. 2005-0042096. That is, the image filtering method using the color coordinate system includes: i) acquiring a Hue channel of the HSI color coordinate system from the color image; ii) acquiring I and Q channels of the YIQ color coordinate system; Combining the I and Q channels to create an H-IQ color coordinate system; i) filtering the image based on the displacement with respect to the H-IQ coordinate system; i) generating corresponding color image position information based on the filtered color image information. And acquiring.

In addition, an image filtering method using a texture may also use the method disclosed in Japanese Patent Laid-Open No. 2005-0042096. That is, the image filtering method using the texture includes i) converting a color image into a gray image, ii) estimating a vertical edge, and iii) dilation to make the edge stand out. do.

Image filtering methods according to the present embodiment as described above includes a process for noise reduction of a video signal and blob labeling for ignoring blobs below a reference size or a reference size of a predetermined license plate. Method is included.

Image filtering methods of the present embodiment use the method disclosed in Japanese Patent Laid-Open No. 2005-0042096, but the present invention is not limited thereto. That is, according to the present invention, if it is possible to select candidate parts which are likely to be located in the license plate among the parts of the captured image, there is no particular limitation on the method.

Meanwhile, FIG. 4 is a diagram illustrating a state in which four candidate parts having a high possibility of locating a license plate in the vehicle area 140 of the captured image are selected using the above method.

Here, since the images of the candidate parts 141, 142, 143, and 144 are all different in size, the candidate parts 141 and 142 may be used for efficient image processing and comparison with the reference plate image. The size of the image of (143) (144) is normalized (step S2). In other words, the sizes of the candidate parts 141, 142, 143, and 144 are the same.

Then, the step of improving the image quality of the images of the candidate portions 141, 142, 143, 144 is performed (step S3). Here, as a method of improving image quality, a contrast enhancement method, a brightness compensation method, or the like may be used.

Next, the signals of the images of the candidate parts 141, 142, 143, and 144 with improved image quality are processed by the frequency use classifier and the spatial information use classifier (steps S4-1 and S4-2).

Here, the frequency using classifier is a device that calculates a similarity value by processing a frequency of an image signal by a method such as a fast fourier transform (FFT) or a discrete cosine transform (DCT).

That is, the reference frequency values of the image signal of the license plate as a reference are calculated and input to the frequency use classifier in advance. Thereafter, the signals of the images of the candidate parts 141, 142, 143, and 144 are processed in a frequency domain, and the similarity values are output by comparing the input frequency with the previously input reference frequency values. Here, the similarity value indicates a higher numerical value as it is similar to the video signal of the reference plate. In the present embodiment, the similarity value is represented as a single numerical value, but the present invention is not limited thereto. That is, according to the present invention, the output similarity value may be represented by a plurality of numerical values.

On the other hand, the spatial information use classifier is a device that calculates the similarity value by processing the image signal for each pixel using a template method.

That is, the gray scale values for each pixel of the license plate image, which are the reference, are calculated and input to the spatial information utilization classifier. Thereafter, the similarity value is output by comparing the gray level values for each pixel of the images of the candidate parts 141, 142, 143, and 144 and the gray level values for each pixel of the pre-input license plate image corresponding thereto. do. Here, the similarity value indicates a higher numerical value as it is similar to the video signal of the reference plate. In the present embodiment, the similarity value is represented as a single numerical value, but the present invention is not limited thereto. That is, according to the present invention, the output similarity value may be represented by a plurality of numerical values.

The spatial information use classifier according to the present embodiment calculates and compares a gray value per pixel of an image of a license plate, but the present invention is not limited thereto. That is, similarity values may be calculated by comparing different properties per pixel in addition to gray levels.

In the present exemplary embodiment, in processing the signals of the images of the candidate parts 141, 142, 143, and 144, the reason for processing through the two classifiers of the frequency classifier and the spatial information classifier is as follows. In other words, due to a change in external lighting or an artificial deformation of the license plate, in some cases, the similarity value calculated from the frequency use classifier may be high, and in some cases, the similarity value calculated from the spatial information use classifier is high. This is because, using both classifiers, the analysis can effectively find the actual license plate area without omission.

Then, similarity values respectively calculated from the frequency use classifier and the spatial information use classifier are normalized and weighted to the similarity values (steps S5-1 and S5-2). That is, the similarity values calculated from the frequency use classifier and the spatial information use classifier are normalized and weighted to be processed to be compared with each other.

First, the similarity value calculated directly from the frequency utilization classifier is normalized. That is, as a criterion of normalization of the similarity value from the frequency use classifier, the normalized similarity value Sim_Fi is obtained by normalizing the minimum value of the normalized similarity value to 0 and the maximum value not to exceed 1. For example, assuming that the minimum value of the similarity value that can come from the frequency use classifier is 0 and the maximum value is 1000, when the similarity value calculated directly from the frequency use classifier is 500, the normalized value of the similarity value Sim_Fi Is normalized to a value of 0.5. That is, in that case, normalization is performed by multiplying the similarity value 500 immediately calculated by the normalization factor 0.001.

In a similar manner, the similarity value calculated from the spatial information use classifier is normalized. That is, as a criterion of normalization of the similarity value from the spatial information use classifier, the normalized similarity value Sim_Si is obtained by normalizing the minimum value of the normalized similarity value to 0 and the maximum value not to exceed 1. For example, assuming that the minimum value of the similarity value that can come from the frequency use classifier is 0 and the maximum value is 400, when the similarity value calculated directly from the frequency use classifier is 100, the normalized value of the similarity value Sim_Si Is normalized to a value of 0.25. That is, in that case, normalization is performed by multiplying the similarity value 100 immediately calculated by the normalization factor 0.0025.

In the present embodiment, as a criterion of normalization of the similarity value from the frequency use classifier and the spatial information use classifier, the normalized similarity value is normalized so that the minimum value is 0 and the maximum value does not exceed 1, but the present invention is not limited thereto. That is, according to the present invention, if the criteria for normalization of the similarity values from the frequency use classifier and the spatial information use classifier are the same, and the comparison of the similarity values from the frequency use classifier and the spatial information use classifier may be easy, There is no special limitation.

After obtaining the normalized similarity values Sim_Fi and Sim_Si as described above, the normalized similarity values Sim_Fi and Sim_Si are weighted.

Here, the weighting method uses a weighting factor α (0 ≦ α ≦ 1), and the weighted similarity values Sim_Ff and Sim_Sf are obtained using the following equations (1) and (2).

In this case, the weighting factor is appropriately corrected because the characteristics of the frequency use classifier and the spatial information use classifier are different. In other words, in order to obtain the weighting factor α, a frequency estimation classifier and a spatial information usage classifier are used to repeatedly obtain the position estimation experiment of the actual license plate. In addition, the weight factor α may be adjusted according to the image quality obtained by the camera, whether or not the image is compressed.

Then, the normalized and weighted similarity values are collected and compared with each other to determine a candidate portion having the largest value among the normalized and weighted similarity values as the license plate position (step S6).

For example, assume that the similarity values of the candidate portions 141, 142, 143, and 144 of FIG. 4 are as follows. Here, the weighting factor α is calculated to be 0.6.

Sim_Fi Sim_Si Sim_Ff Sim_Sf Candidate Part 141 0.3 0.5 0.18 0.2 Candidate Part 142 0.5 0.8 0.3 0.32 Candidate Part 143 0.4 0.6 0.24 0.24 Candidate Part 144 0.2 0.3 0.12 0.12

Here, the largest value of the Sim_Ff and Sim_Sf values of the candidate parts 141, 142, 143, and 144 is 0.32, which is the Sim_Sf value of the candidate part 142. Thus, the candidate portion 142 is determined to be the license plate position of the vehicle.

Here, in order to determine the position of the license plate of the vehicle, it is not determined by adding or multiplying the value of Sim_Ff and the value of Sim_Sf of each candidate portion, but having the largest maximum value through the values of Sim_Ff and Sim_Sf is selected. The candidate portion having the maximum value is determined as the license plate area.

As described above, the image signal of the license plate image is distorted due to deterioration of external lighting, weather conditions, etc., so that the similarity value calculated from either the frequency utilization classifier or the spatial information usage classifier may not be accurately obtained. The similarity value calculated in FIG. 3 allows the vehicle license plate position to be determined.

For example, in the photographed image of FIG. 4, when the image signal is distorted due to heavy rain and the processing of the spatial information use classifier is incorrect, it is assumed that the similarity value is calculated as shown in Table 2 below. The value of the weighting factor α at this time is also 0.6 as in the case of the above-described example.

Sim_Fi Sim_Si Sim_Ff Sim_Sf Candidate Part 141 0.3 0.2 0.18 0.08 Candidate Part 142 0.5 0.1 0.3 0.04 Candidate Part 143 0.4 0.2 0.24 0.08 Candidate Part 144 0.2 0.1 0.12 0.04

As shown in Table 2 above, the function of the spatial information utilization classifier is not properly exhibited, so the similarity value of Sim_Sf is low overall, and the value of Sim_Sf of the candidate part 142 which is the actual license plate area is determined by the candidate part 143. It becomes smaller than the value of Sim_Sf. However, according to the determination method of the present embodiment, since the candidate portion 142 having the largest value 0.3 is selected throughout the values of Sim_Ff and Sim_Sf, the candidate portion 142 is determined as the license plate area even in this case. Thus, according to this embodiment, even if the similarity value calculated from the spatial information use classifier is not accurate, the vehicle license plate position can be determined accurately.

In addition, as a comparative example, the case where the function of the spatial information use classifier is not properly operated has been described, but the above description may be applied to the case where the function of the frequency domain classifier is not properly operated.

In the above manner, after the candidate portion determined by the license plate is determined, a process of identifying a character in the area of the candidate portion 142 determined by the license plate is performed, and such a method may be a known character recognition method. .

According to this embodiment, the license plate recognition unit 134 performs both the operation of estimating the position of the vehicle license plate and the operation of recognizing the characters of the license plate whose position is estimated, but the present invention is not limited thereto. That is, according to the present invention, a separate license plate position estimation apparatus for performing only the license plate position estimation method described above may be configured. When configured separately, the hardware type of the license plate position estimation device may be an electronic circuit, or may be a type such as a ROM in which software of the license plate position estimation method is mounted. On the other hand, when a separate license plate position estimation device is configured, a separate license plate character recognition device for recognizing characters on the license plate is required.

As described above, after estimating the position of the license plate and recognizing the letters of the license plate, the enforcement unit 136 performs the enforcement operation as described above.

As described above, according to the license plate position estimation method of the present embodiment, the similarity values calculated from the frequency classifier and the spatial information classifier are normalized to be compared with each other, and the candidate part having the maximum value among the normalized similarity values is licensed. By determining the area, even if the similarity value calculated from either the frequency use classifier or the spatial information use classifier is not accurate, the exact license plate position can be determined using the similarity value calculated in the other. As a result, the position of the vehicle license plate can be accurately determined from the captured image, thereby enabling efficient parking control.

As described above, according to the present invention, there is an effect of accurately estimating the position of the license plate of the vehicle in the image acquired from the imaging device.

While the present invention has been described with reference to exemplary embodiments, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims. Accordingly, the true scope of the present invention should be determined by the technical idea of the appended claims.

Claims (10)

In a vehicle license plate position estimation method of a parking lot monitoring system having an imaging device, (a) selecting a plurality of candidate portions having a high possibility of the vehicle license plate being located among the portions of the image captured by the imaging device; (b) calculating, by the frequency use classifier and the spatial information use classifier, the similarity values using the respective video signals of the selected candidate parts; (c) normalizing and weighting the similarity values calculated by the frequency use classifier and the spatial information use classifier for mutual comparison; And (d) determining the candidate portion having the largest value among the normalized and weighted similarity values as a license plate position by collecting and comparing the normalized and weighted similarity values to each other; Location estimation method. The method of claim 1, And (b) applying an image filtering method using a color coordinate system to the captured images in order to select the candidate parts. The method of claim 1, And (b) applying the image filtering method using the texture of the photographed image to select the candidate portions. The method according to claim 2 or 3, The image filtering methods of the vehicle license plate position estimation method using a blob labeling method that ignores blobs of a predetermined size or less than a predetermined size. The method of claim 1, Between steps (a) and (b), (e) normalizing the size of an image of the selected candidate portions; And and (f) improving image quality of the images passed through step (e). The method of claim 5, The improving of the image quality in the step (f) is a vehicle license plate position estimation method using a contrast enhancement method. The method of claim 5, The step of improving the image quality in the step (f) is a vehicle license plate position estimation method using a brightness compensation method. The method of claim 1, In the step (b), the frequency utilization classifier calculates a similarity value by using the video signals of the selected candidate parts. And calculating a similarity value by comparing frequency values of a video signal of a reference plate pre-input with the frequency values of the video signals of the selected candidate parts. The method of claim 1, In the step (b), the spatial information use classifier calculates a similarity value by using the respective video signals of the selected candidate parts. And calculating a similarity value by comparing grayscale values of pixels of an image of a reference license plate previously input with grayscale values of pixels of an image of the selected candidate parts. delete

Images