WO2015184899A1

WO2015184899A1 - Method and device for recognizing license plate of vehicle

Info

Publication number: WO2015184899A1
Application number: PCT/CN2015/074792
Authority: WO
Inventors: 王勃飞; 郑成建; 夏珍; 罗艳红; 黄晓红
Original assignee: 中兴通讯股份有限公司
Priority date: 2014-07-23
Filing date: 2015-03-20
Publication date: 2015-12-10
Also published as: CN105303153A

Abstract

A method and device for recognizing a license plate of a vehicle. The method comprises the steps of: recognizing image frames of a license plate in a plurality of continuous positions within a picture acquisition range; and respectively obtaining recognition data of the license plate in the plurality of continuous positions; and making statistics about the recognition data of the license plate in the plurality of continuous positions, to obtain a recognition result of the license plate. The device comprises: a recognition data acquisition module configured to recognize image frames of a license plate in a plurality of continuous positions within a picture acquisition range; and respectively obtain recognition data of the license plate in the plurality of continuous positions; and a statistics module configured to count the recognition data of the license plate in the plurality of positions, to obtain a recognition result of the license plate. The method and device can increase the accuracy of license plate recognition for a vehicle in the case where an image collection effect cannot be stably guaranteed due to external environment and other reasons.

Description

Vehicle license plate recognition method and device

Technical field

The present invention relates to the field of image technologies, and in particular, to a vehicle license plate recognition method and apparatus.

Background technique

With people's life development and technological advancement, license plate recognition has gradually become an important part of the intelligent transportation system. In the prior art, the active license plate recognition technology for the bayonet has matured, and the technology can extract and recognize the moving license plate from the complex background, through license plate extraction, image preprocessing, feature extraction, license plate characters. Identification and other techniques to identify vehicle grades. This kind of license plate recognition technology is highly commercial, and mainstream license plate recognition manufacturers can achieve more than 90% recognition accuracy. However, with the continuous increase of traffic volume in the city and complicated traffic conditions, the active license plate recognition technology for bayonet has been unable to meet the needs of practical applications, in order to quickly correct traffic violations, quickly detect traffic accidents and escape and motor vehicle robbery. The case is crucial for passive license plate recognition technology for moving vehicle license plate recognition. Then, influenced by many factors, the recognition rate of such technologies is still very low, and the measured indicators of mainstream manufacturers show that the correct rate is less than 80%.

The license plate recognition technology in the prior art faces new challenges with the development of urban traffic. Due to various uncontrollable factors in the environment of the vehicle, such as occlusion due to positional changes, image distortion, etc., the quality of the captured image such as the bayonet cannot be stably guaranteed, thereby affecting the accuracy of the license plate recognition.

Summary of the invention

In view of this, an embodiment of the present invention provides a vehicle license plate recognition method and apparatus. The accuracy of the vehicle license plate recognition can be improved under the condition that the image acquisition effect due to the external environment and the like cannot be stably ensured.

A vehicle license plate recognition method according to an embodiment of the present invention, comprising the following steps:

Identifying image frames of the plurality of consecutive positions of the license plate in the screen acquisition range; and obtaining identification data of the license plate when the plurality of consecutive positions are respectively obtained;

When the plurality of locations are counted, the identification data of the license plate obtains the recognition result of the license plate.

Optionally, the step of identifying the image frames of the plurality of consecutive positions of the license plate in the screen acquisition range and obtaining the identification data of the license plate in the plurality of consecutive positions respectively comprises:

Positioning the license plate in an image frame to obtain a corresponding area of the license plate in the image frame, and parsing a corresponding area of the license plate in the image frame to obtain a character recognition result of the license plate picture;

The character recognition result is output as license plate identification data.

Optionally, before identifying the image frames of the plurality of locations within the screen acquisition range and obtaining the license plate identification data when the plurality of locations are respectively obtained, the method further includes:

Identifying the first image frame of the license plate entering a screen acquisition range;

Recognizing that the license plate leaves the second image frame of the screen acquisition range;

All image frames between the first image frame and the second image frame are used as image frames of the license plate at a plurality of consecutive positions within the screen acquisition range.

Optionally, the step of identifying the image frame of the license plate entering the screen acquisition range comprises:

Determining whether a license plate in an image frame has an associated license plate in a consecutively set number of image frames in front of it;

Determining that the license plate in the one image frame does not have an associated license plate in a consecutively set number of image frames in front of the license plate;

Taking the one image frame as a first image frame;

The associated license plate refers to a license plate that appears in two adjacent image frames and has a similarity equal to or greater than a set value.

Optionally, after determining that the license plate in the one image frame does not have an associated license plate in a consecutively set number of image frames in front of the image frame, the method further includes:

Determining whether the license plate in the one image frame has an associated license plate in the currently recorded unrecognized license plate set;

When the license plate in the one image frame does not have an associated license plate in a consecutively set number of image frames, the step of using the one image frame as the first image frame is entered.

Optionally, the step of identifying that the license plate leaves the second image frame of the screen acquisition range comprises:

And adopting the first image frame and the subsequent time-series adjacent image frame as the target image frame;

Determining whether the license plate in the current target image frame has an associated license plate in a continuously set number of image frames behind it;

Determining, when the license plate in the current target image frame does not have an associated license plate in a consecutively set number of image frames, the target image frame is used as the second image frame;

Determining that, when the license plate in the current target image frame has an associated license plate in a consecutive set number of image frames, the next target image frame of the current target image frame is used as the target image frame, and returns to determine the current target image. The step of whether the license plate in the frame has an associated license plate in a consecutive number of image frames behind it;

Optionally, when the distance between the center of the two license plates respectively appearing in two adjacent image frames is less than the set distance threshold and the similarity ratio of the license plate character recognition result is greater than the set similarity ratio threshold, the two phases The two license plates in the adjacent image frame are associated license plates.

Further, an embodiment of the present invention provides a vehicle license plate recognition device, including:

a recognition data acquisition module: configured to identify image frames of a plurality of consecutive positions of the license plate within the screen acquisition range; and obtain identification data of the license plate when the plurality of consecutive positions are respectively obtained;

The statistics module: when the plurality of locations are counted, the identification data of the license plate is used to obtain the recognition result of the license plate.

Optionally, the identification data acquisition module includes:

a character recognition unit configured to: position the license plate in an image frame to obtain a corresponding area of the license plate in the image frame, and parse a corresponding area of the license plate in the image frame to obtain a character recognition result of the license plate picture;

The license plate recognition data output unit is configured to output the character recognition result as license plate identification data.

Optionally, the device further includes:

a first image frame identification module: configured to identify a first image frame in which the license plate enters a screen acquisition range;

a second image frame identification module: configured to identify the second image frame of the license plate leaving the screen acquisition range;

The image frame extraction module is configured to set all the image frames between the first image frame and the second image frame as image frames of the plurality of consecutive positions of the license plate within the image acquisition range.

Optionally, the first image frame identification module includes:

a first association determining unit: configured to determine whether an associated license plate exists in a consecutively set number of image frames of a license plate in an image frame;

a first determination result determining unit: configured to determine that the license plate in the one image frame does not have an associated license plate in a consecutively set number of image frames in front of the license plate;

a first image frame determining unit: configured to use the one image frame as a first image frame;

Optionally, the first image frame identification module further includes:

Unrecognized record association judging unit: configured to determine whether the license plate in the one image frame has an associated license plate in the currently recorded unrecognized license plate set; and set to continuously set the license plate in the one image frame before The first image frame determining unit is triggered when there is no associated license plate in the number of image frames.

Optionally, the second image frame identification module includes:

a target image frame determining unit: configured to use the temporally adjacent image frame of the first image frame as a target image frame;

a second association determining unit: configured to determine whether an associated license plate exists in a continuously set number of image frames of the license plate in the current target image frame;

a second image frame determining unit: configured to determine that the target image frame is the second image frame when the license plate in the current target image frame does not have an associated license plate in a consecutively set number of image frames; In order to determine that the license plate in the current target image frame has an associated license plate in a consecutive set number of image frames, the next target image frame is used as the target image frame, and the second association is triggered. Judging unit

It can be seen from the above that the vehicle license plate recognition method and device provided by the embodiments of the present invention are directed to the problem that the license plate recognition result is adversely affected due to the unstable image acquisition condition in the actual application process, and the vehicle enters the screen. In the continuous process of obtaining the range, the image frames of multiple positions appearing successively are processed to obtain the identification data, and the identification data is statistically obtained to obtain the license plate recognition result, thereby improving the accurate recognition of the license plate under the influence of environmental factors. Probability. The vehicle license plate recognition method and apparatus according to the embodiment of the present invention can perform high-accuracy determination of the state in which the license plate enters and leaves the screen acquisition range by association recognition. The license plate vehicle identification method and apparatus provided by the present invention and its embodiments are particularly suitable for a mobile vehicle to adapt to the current traffic development trend.

DRAWINGS

1 is a schematic flow chart of a vehicle license plate recognition method according to an embodiment of the present invention;

2 is a schematic diagram of steps involved in recognizing a license plate before an image frame of a plurality of consecutive positions within a screen acquisition range according to an embodiment of the present invention;

3 is a schematic structural view of a vehicle license plate recognition device according to an embodiment of the present invention.

detailed description

In order to provide an effective implementation, the present invention provides the following embodiments, which are described below in conjunction with the accompanying drawings.

The embodiment of the invention firstly provides a vehicle license plate recognition method, which comprises the steps shown in FIG. 1 :

Step 101: Identify image frames of a plurality of consecutive positions of the license plate in the screen acquisition range; and obtain identification data of the license plate when the plurality of consecutive positions are respectively obtained;

Step 102: When the plurality of locations are counted, the identification data of the license plate obtains the recognition result of the license plate.

As can be seen from the above, the vehicle license plate recognition method provided by the embodiment of the present invention identifies an image frame of a plurality of consecutive positions within the screen acquisition range of the license plate and obtains corresponding license plate identification data, so that when the license plate is within the screen acquisition range. When most of the image frames are occluded, or for other reasons, most of the image frames in the image acquisition range are not recognized, the probability that the license plate cannot be recognized is reduced. In addition, by counting the license plate identification data at multiple locations, the license plate recognition error rate can be reduced and the recognition accuracy can be improved.

The image frames of the plurality of consecutive positions may be image frames that are continuously played in the video, or may be image frames that are acquired by the photographing device at consecutively set time intervals.

In the embodiment of the present invention, the license plate is in a plurality of consecutive positions within the screen acquisition range, and refers to a plurality of positions in which the license plate appears successively in a complete process of entering the screen acquisition range to leaving the screen acquisition range.

In some preferred embodiments of the present invention, the step of identifying the image frames of the plurality of consecutive positions within the screen acquisition range of the license plate and obtaining the identification data of the license plate at the plurality of consecutive positions respectively comprises:

Step 1011: Positioning the license plate in an image frame to obtain a corresponding area of the license plate in the image frame, and parsing a corresponding area of the license plate in the image frame to obtain a character recognition result of the license plate picture;

Step 1012: Output the character recognition result as license plate identification data.

When the image frame is from a video, a decoded video image frame is read in, and then the image frame is processed in accordance with step 1011.

In some embodiments of the present invention, in the above step 1011, the step of parsing the corresponding area of the license plate in the image frame to obtain the character recognition result of the license plate picture may be implemented by the following steps 1031-1032:

Step 1031: Positioning the license plate in the currently read image frame.

Alternatively, the license plate positioning may be performed by combining the grayscale projection information and the texture information method of the license plate area. Firstly, the image is gray-scale transformed, and the edge image is generated by edge detection using Sobel operator. Then, the horizontal and vertical projection images of the edge image are calculated and the possible license plate area is determined according to the peak and valley information of the projected image. Finally, according to The license plate character area is rich in texture and further determines and cuts the possible license plate area to determine the final license plate area.

Step 1032: The license plate picture character is divided.

The task of character segmentation is to cut each character in the license plate image from the entire image into a single character. Common methods of segmentation include projection-based segmentation, matching-based segmentation, and clustering-based segmentation.

The projection-based segmentation includes simple vertical projection segmentation, upper and lower edge projection segmentation, and the like.

The matching based segmentation includes template matching segmentation.

The cluster-based segmentation divides the characters in the license plate image according to the principle that the pixels belonging to the same character form a connected domain, and combines the prior knowledge of the license plate height and the fixed proportional relationship of the spacing.

When step 1032 is executed, first, the license plate acquisition image may be tilted to some extent due to problems such as the shooting angle, so the license plate image should be corrected. In a specific embodiment, the license plate rotation angle can be estimated according to the edge gradient information of the license plate region, and then the license plate is corrected to the horizontal state by affine transformation; then, the algorithm initially determines the license plate type according to the license plate background color, and combines the character region. The peak-to-valley information pair projected in the vertical direction separates each license plate character.

Step 1033: License plate character recognition.

In the above step 1033, features are extracted and identified for the characters segmented in step 1031. Common features include direction line features, mesh features, gradient features, and the like. Commonly used recognition methods include template matching method, linear discriminating method, SVM (Support Vector Machine) and so on.

In a specific embodiment, the segmented character extraction direction line feature is used to identify the license plate characters: First, a common series of preprocessing operations including smoothing, contrast enhancement, position normalization, size normalization, and binarization , to obtain a normalized grayscale image or binarization map of size 64x64 pixels. In this embodiment, the character stroke linear feature is extracted, and the Mahalanobis distance between it and all the character template features to be recognized is calculated, and the character corresponding to the closest template feature is selected as the result of the character recognition.

After the character recognition result is output as the license plate identification data, the character recognition result can be statistically analyzed, that is, the recognition result with the highest probability of occurrence in the recognition result of the same character position of the license plate is selected as the recognition of the character position on the license plate. result. For example, by recognizing seven image frames of seven successive positions of the license plate within the screen acquisition range; and obtaining identification data of the license plate at the seven consecutive positions, respectively. The identification data of the first to seventh image frames are: Beijing BEQ7C2, Beijing BF02C2, expensive BE0702, Yue BE07C2, Yue BF07C2, Yue BE07C2, Yue BE07C2, Yue BE07C2. The character recognition results of the seven character positions are respectively counted; if the first character position is identified as "Beijing" 2 times, the recognition as "expensive" has 1 time, and the recognition as "Yue" has 4 times, it is considered that The final recognition result of the position is "Yue"; similarly, the recognition results of the second to seventh character positions are: "B", "E", "0", "7", "C", "2".

In a preferred embodiment, if there are multiple recognition results with the highest probability of occurrence in the same character position, the plurality of recognition results are recorded for selection.

In some embodiments of the present invention, prior to identifying the image frames of the plurality of locations within the screen acquisition range and obtaining the license plate identification data at the plurality of locations, the process illustrated in FIG. 2 is also included:

Step 201: Identify that the license plate enters a first image frame of a screen acquisition range;

Step 202: Identify that the license plate leaves the second image frame of the screen acquisition range;

Step 203: Use all image frames between the first image frame and the second image frame as image frames of the plurality of consecutive positions in the screen acquisition range of the license plate.

In the above embodiment of the present invention, the image frames acquired by the license plate belonging to the same vehicle entering the screen acquisition range and the exit screen acquisition range are used as the image frames of the license plate at a plurality of consecutive positions within the screen acquisition range. Tracking the license plate when it is in the screen acquisition range, to the greatest extent, ensuring that at least one image frame of the license plate can be used as an image object recognized by the license plate, and can overcome a certain frame or a certain number during the running of the vehicle. The frame acquisition image can not identify or identify the wrong interference, and solves the problem that the image quality of the license plate acquisition image is difficult to be stabilized when the license plate recognition of the moving vehicle is simple and effective.

In some embodiments of the invention, the step of identifying the license plate entering the image frame of the screen acquisition range comprises:

Step 2011: determining whether the license plate in an image frame has an associated license plate in a consecutively set number of image frames in front of the license plate;

Step 2012: determining that the license plate in the one image frame does not have an associated license plate in a consecutively set number of image frames in front of the license plate;

Step 2013: using the image frame as a first image frame;

In a specific embodiment of the present invention, when processing each frame image, all license plate recognition results of the previous time series image frame are associated with all license plate recognition results of the current image frame frame, and the determination method is: if the previous timing is adjacent If the i-th license plate of the image frame and the j-th license plate similarity of the current image frame reach the set condition, it is considered that the i-th license plate of the preceding sequential adjacent image frame and the j-th license plate of the current image frame can be associated.

In the initialization state, all the license plates in the time series image frame in front of the first image frame are zero.

For an image frame, if a license plate in the preceding adjacent time-series image frame does not have an associated license plate in the subsequent adjacent time-series image frame, the license plate is considered to be unrecognizable at present, and the license plate is added to the unrecognizable license plate set.

In some cases, for a license plate k in an image frame, if there is no associated license plate in the preceding adjacent time-series image frame, the image frame may be considered as the first image frame in which the license plate k enters the screen acquisition range.

In other cases, for the license plate m in an image frame, when there is no associated license plate in the preceding adjacent time-series image frame, it is possible that the license plate m is occluded in consecutive n-1 image frames in front of the image frame or Not being clearly captured, but appearing in the nth image frame before the image frame, then the image frame is the first image frame in which the license plate m enters the screen acquisition range and the second image frame in which the license plate m leaves the screen acquisition range. Image frame between.

Therefore, after step 2012, it also includes:

Step 2014: determining whether the license plate in the one image frame has an associated license plate in the currently recorded unrecognizable license plate set;

Step 2015: When the license plate in the one image frame does not have an associated license plate in a consecutively set number of image frames, the process proceeds to step 2013.

In the actual processing, in some cases, for the license plate h in an image frame, if there is no associated license plate in the sequence of adjacent image frames behind the image frame, the image frame may be considered as the license plate h leaving the screen. The second image frame of the range.

In other cases, for the license plate g in an image frame, when there is no associated license plate in the subsequent adjacent time-series image frames, it is possible that the license plate g is occluded in consecutive n-1 image frames behind the image frame or Without being clearly captured, and appearing again in the nth image frame of the image frame, the image frame is the first image frame in which the license plate g enters the screen acquisition range and the second image frame in which the license plate g leaves the screen acquisition range. Image frame between.

In some embodiments of the invention, still referring to FIG. 2, step 202 includes:

Step 2021: The subsequent image frame of the first image frame is used as the target image frame.

Step 2022: Determine whether the license plate in the current target image frame has an associated license plate in a consecutively set number of image frames behind it;

Step 2023: Determine that the target image frame is used as the second image frame when the license plate in the current target image frame does not have an associated license plate in a consecutively set number of image frames.

Step 2024: When it is determined that the license plate in the current target image frame has an associated license plate in a consecutive set number of image frames, the next target image frame of the current target image frame is used as the target image frame; and the judgment is returned. a step of whether the license plate in the current target image frame has an associated license plate in a continuously set number of image frames behind it;

In some embodiments of the present invention, for a certain image frame, if a certain license plate in the preceding adjacent time-series image frame does not have an associated license plate in a subsequent adjacent time-series image frame, the license plate is considered to be unrecognizable at present. And add the license plate to the unrecognizable license plate collection. In a specific embodiment of the present invention, the above step 2022 can be implemented by the following process:

Step 301: Determine whether the license plate in the current target image frame has an associated license plate in the first image frame behind it. If not, add the license plate to the unrecognizable license plate set, record the number of unrecognizable numbers as 1, and proceed to the next step. If yes, proceed to step 2024 above.

Step 302: Determine whether the unrecognized number of times reaches the set value, and if yes, proceed to step 2023; if not, proceed to the next step.

Step 303: Determine whether the associated license plate exists in the next image frame of the license plate in the current target image frame. If not, increase the number of unrecognized times by one, and return to step 302; if yes, proceed to step 2024.

In some embodiments of the present invention, when the distance between the center of the two license plates respectively appearing in two adjacent image frames is less than the set distance threshold and the similarity ratio of the license plate character recognition result is greater than the set similarity ratio threshold, Two of the two adjacent image frames are associated license plates.

In a specific embodiment, if the distance between the center point of the area where the a-th license plate of the last processing frame and the b-th license plate of the current processing frame are located is less than 1/4 of the diagonal length of the image, and no less than 5 characters are recognized. Similarly, we believe that the a-th license plate of the previous processing frame and the b-th license plate of the current processing frame can be associated.

It can be seen from the above that the vehicle license plate recognition method provided by the present invention and the embodiments thereof can solve the problem that the recognition result is inaccurate or the recognition result cannot be generated due to insufficient clarity or occlusion of the captured image frame of the vehicle. . At the same time, the accuracy of recognition has been improved.

Further, an embodiment of the present invention provides a vehicle license plate recognition device, and the structure is as shown in FIG. 3, including:

In some embodiments of the present invention, the identification data acquisition module includes:

In some embodiments of the present invention, still referring to FIG. 3, the apparatus further includes:

In some embodiments of the present invention, the first image frame identification module includes:

In some embodiments of the present invention, the first image frame identification module further includes:

In some embodiments of the present invention, the second image frame identification module includes:

In some embodiments of the present invention, when the center distance of the area where the two license plates respectively appear in two adjacent image frames are less than the set distance threshold and the similarity ratio of the license plate character recognition result is greater than the set similarity ratio threshold, Two of the two adjacent image frames are associated license plates.

It can be seen from the above that the vehicle license plate recognition method and device provided by the embodiments of the present invention are directed to the problem that the license plate recognition result is adversely affected due to the unstable image acquisition condition in the actual application process, and the vehicle enters the screen. In the continuous process of obtaining the range, the image frames of multiple positions appearing successively are processed to obtain the identification data, and the identification data is statistically obtained to obtain the license plate recognition result, thereby improving the accurate recognition of the license plate under the influence of environmental factors. Probability. The vehicle license plate recognition method and apparatus according to the embodiment of the present invention can perform high-accuracy determination of the state in which the license plate enters and leaves the screen acquisition range by association recognition.

Industrial Applicability: The vehicle license plate recognition method and apparatus according to the embodiment of the present invention can perform high-accuracy determination of the state in which the license plate enters and leaves the screen acquisition range by association recognition. The license plate vehicle identification method and apparatus provided by the present invention and its embodiments are particularly suitable for a mobile vehicle to adapt to the current traffic development trend.

It is to be understood that the various embodiments of the present invention are intended to illustrate and explain the invention. And in the case of no conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.

It is apparent that those skilled in the art can make various modifications and variations to the invention without departing from the spirit and scope of the invention. Thus, it is intended that the present invention cover the modifications and modifications of the invention

Claims

A vehicle license plate recognition method includes:

Identifying image frames of the plurality of consecutive positions of the license plate in the screen acquisition range; and obtaining identification data of the license plate when the plurality of consecutive positions are respectively obtained;

When the plurality of locations are counted, the identification data of the license plate obtains the recognition result of the license plate.
The method according to claim 1, wherein the step of identifying the image frames of the plurality of consecutive positions of the license plate within the screen acquisition range and obtaining the identification data of the license plate at the plurality of consecutive positions respectively comprises:

Positioning the license plate in an image frame to obtain a corresponding area of the license plate in the image frame, and parsing a corresponding area of the license plate in the image frame to obtain a character recognition result of the license plate picture;

The character recognition result is output as license plate identification data.
The method according to claim 1, wherein before identifying the image frames of the plurality of locations within the screen acquisition range and obtaining the license plate identification data at the plurality of positions, the method further comprises:

Identifying the first image frame of the license plate entering a screen acquisition range;

Recognizing that the license plate leaves the second image frame of the screen acquisition range;

All image frames between the first image frame and the second image frame are used as image frames of the license plate at a plurality of consecutive positions within the screen acquisition range.
The method of claim 3, wherein the step of identifying the license plate entering the image frame of the screen acquisition range comprises:

Determining whether a license plate in an image frame has an associated license plate in a consecutively set number of image frames in front of it;

Determining that the license plate in the one image frame does not have an associated license plate in a consecutively set number of image frames in front of the license plate;

Taking the one image frame as a first image frame;

The associated license plate refers to a license plate that appears in two adjacent image frames and has a similarity equal to or greater than a set value.
The method according to claim 4, wherein after determining that the license plate in the one image frame has no associated license plate in a consecutively set number of image frames in front of the image frame, the method further comprises:

Determining whether the license plate in the one image frame has an associated license plate in the currently recorded unrecognized license plate set;

When the license plate in the one image frame does not have an associated license plate in a consecutively set number of image frames, the step of using the one image frame as the first image frame is entered.
The method of claim 3, wherein the step of identifying the second image frame of the license plate leaving the screen acquisition range comprises:

And adopting the first image frame and the subsequent time-series adjacent image frame as the target image frame;

Determining whether the license plate in the current target image frame has an associated license plate in a continuously set number of image frames behind it;

Determining, when the license plate in the current target image frame does not have an associated license plate in a consecutively set number of image frames, the target image frame is used as the second image frame;

Determining that, when the license plate in the current target image frame has an associated license plate in a consecutive set number of image frames, the next target image frame of the current target image frame is used as the target image frame, and returns to determine the current target image. The step of whether the license plate in the frame has an associated license plate in a consecutive number of image frames behind it;

The associated license plate refers to a license plate that appears in two adjacent image frames and has a similarity equal to or greater than a set value.
The method according to any one of claims 4-6, wherein the center distance of the area where the two license plates respectively appear in the two adjacent image frames is smaller than the set distance threshold and the similarity ratio of the license plate character recognition result is greater than When the similarity ratio threshold is set, two of the two adjacent image frames are associated license plates.
A vehicle license plate recognition device includes:

a recognition data acquisition module: configured to identify image frames of a plurality of consecutive positions of the license plate within the screen acquisition range; and obtain identification data of the license plate when the plurality of consecutive positions are respectively obtained;

The statistics module: when the plurality of locations are counted, the identification data of the license plate is used to obtain the recognition result of the license plate.
The apparatus of claim 8, wherein the identification data acquisition module comprises:

a character recognition unit configured to: position the license plate in an image frame to obtain a corresponding area of the license plate in the image frame, and parse a corresponding area of the license plate in the image frame to obtain a character recognition result of the license plate picture;

The license plate recognition data output unit is configured to output the character recognition result as license plate identification data.
The apparatus of claim 8 wherein said apparatus further comprises:

a first image frame identification module: configured to identify a first image frame in which the license plate enters a screen acquisition range;

a second image frame identification module: configured to identify the second image frame of the license plate leaving the screen acquisition range;

The image frame extraction module is configured to set all the image frames between the first image frame and the second image frame as image frames of the plurality of consecutive positions of the license plate within the image acquisition range.
The apparatus of claim 10 wherein the first image frame identification module comprises:

a first association determining unit: configured to determine whether an associated license plate exists in a consecutively set number of image frames of a license plate in an image frame;

a first determination result determining unit: configured to determine that the license plate in the one image frame does not have an associated license plate in a consecutively set number of image frames in front of the license plate;

a first image frame determining unit: configured to use the one image frame as a first image frame;

The associated license plate refers to a license plate that appears in two adjacent image frames and has a similarity equal to or greater than a set value.
The apparatus of claim 11, wherein the first image frame identification module further comprises:

Unrecognized record association judging unit: configured to determine whether the license plate in the one image frame has an associated license plate in the currently recorded unrecognized license plate set; and set to continuously set the license plate in the one image frame before The first image frame determining unit is triggered when there is no associated license plate in the number of image frames.
The apparatus of claim 10 wherein the second image frame identification module comprises:

a target image frame determining unit: configured to use the temporally adjacent image frame of the first image frame as a target image frame;

a second association determining unit: configured to determine whether an associated license plate exists in a continuously set number of image frames of the license plate in the current target image frame;

a second image frame determining unit: configured to determine that the target image frame is the second image frame when the license plate in the current target image frame does not have an associated license plate in a consecutively set number of image frames; In order to determine that the license plate in the current target image frame has an associated license plate in a consecutive set number of image frames, the next target image frame is used as the target image frame, and the second association is triggered. Judging unit

The associated license plate refers to a license plate that appears in two adjacent image frames and has a similarity equal to or greater than a set value.
The apparatus according to any one of claims 11 to 13, wherein when the distance between the center of the two license plates respectively appearing in the two adjacent image frames is less than the set distance threshold and the similarity ratio of the license plate character recognition result is greater than When the similarity ratio threshold is set, two of the two adjacent image frames are associated license plates.