TWM592542U - License plate recognition system - Google Patents

Abstract

A license plate recognition system includes an image capturing unit and a processing unit. The processing unit obtains a character detection result according to one of the images to be processed captured by the image capturing unit, and when it is determined that the character detection result indicates that the image to be processed contains a plurality of characters to be processed In the meta block, a representative point corresponding to each character block to be processed is obtained, the processing unit obtains a straight line closest to each representative point, the processing unit obtains a correction angle according to the straight line and a reference line, The processing unit corrects the waiting character block according to the correction angle to obtain a plurality of correction character blocks. The processing unit uses the image character recognition model to recognize the correction character block according to the correction character block The license plate characters corresponding to the correction character blocks.

Description

License Plate Recognition System

The present invention relates to an image recognition system, especially an image recognition system applied to license plates.

The research on the image recognition technology used for license plate recognition has been carried out for a long time, and there are various application facilities related to license plate recognition in the market (for example: parking lot). Although there are many license plate recognition technologies on the market, the training data and the ease of construction required for how to process or correct images are still different. Therefore, there is still a need for manufacturers to provide other brand-new license plate recognition technology construction methods to provide enterprises or merchants to choose and use according to their own needs.

Based on the above, a brand new license plate recognition system is provided here, which is the primary goal of this creation.

Therefore, the purpose of this new model is to provide a brand new license plate recognition system.

Therefore, the new vehicle license plate recognition system includes an image capturing unit and a processing unit electrically connected to the image capturing unit.

Wherein, the processing unit uses a first object for detecting a character block corresponding to each license plate character in the image according to an image to be processed captured by the image capturing unit A model to obtain a character detection result corresponding to the image to be processed, and the processing unit determines whether the character detection result indicates that the image to be processed includes a plurality of characters corresponding to a plurality of license plate characters of a license plate Processing character blocks, when the processing unit determines that the character detection result indicates that the to-be-processed image includes the waiting character block, for each character to be processed in the character detection result Block, the processing unit obtains a representative point corresponding to the character block to be processed, the processing unit obtains a straight line closest to each representative point, and the processing unit obtains a representative point according to the straight line and a preset reference line Correction angle. For each character block to be processed, the processing unit corrects the character block to be processed according to the correction angle to obtain a correction character block. For each correction character block, The processing unit recognizes the license plate character corresponding to the corrected character block using an image character recognition model according to the corrected character block.

The effect of the present invention lies in that: by using the processing unit, the first object detection model is used to obtain the character detection result corresponding to the image to be processed, and then the waiting indicated by the character detection result Processing character blocks, obtaining the straight line corresponding to the representative point corresponding to each character block to be processed, and after obtaining the correction angle according to the straight line, correcting each character block to be processed to the The correction character block finally recognizes the license plate character corresponding to each correction character block.

Referring to FIG. 1, an embodiment of the new license plate recognition system 1 includes an electronic device 2 and a server 3 connected to the electronic device 2 via a communication network 200.

The electronic device 2 includes a first communication module 21 connected to the communication network 200, a first storage module 22, a first image capture module 23, and an electrically connected first communication module 21, the The first storage module 22 and the first processing module 24 of the first image capture module 23. In particular, the first image capture module 22 is included in an image capture unit.

The server 3 includes a second communication module 31 connected to the communication network 200, a second storage module 32, and a second electrically connected to the second communication module 31 and the second storage module 32 Processing module 33. In particular, the first processing module 24 and the second processing module 33 are included in a processing unit.

The first storage module 22 or the second storage module 32 stores a first object detection model for detecting a character block corresponding to each license plate character in the image, and one for detecting A second object detection model for a vehicle block corresponding to a vehicle in the image, and a third object detection model for detecting a license plate block corresponding to a license plate in the image. It is worth noting that the first object detection model, the second object detection model and the third object detection model are all multi-level block convolutional neural networks (R-CNN, Region with Convolutional Neural Network). Wherein, the first object detection model is trained using multiple first training data, and each first training data includes a license plate image and a block indicating the block where each license plate character in the license plate image is located First location information. The second object detection model is trained using a plurality of second training data, and each second training data includes an image and second position information indicating the block where the vehicle in the image is located. The third object detection model is trained using multiple third training data, and each third training data includes a vehicle image and third position information indicating the block where the license plate in the vehicle image is located.

In this embodiment, the implementation of the electronic device 2 is, for example, an electronic device with a camera function, but it is not limited thereto.

In this embodiment, the implementation of the server 3 is, for example, a personal computer, a server, or a cloud host, but not limited to this.

Referring to FIG. 2, the operation details of the components of the electronic device 2 and the server 3 will be described below by performing a license plate recognition method by the embodiment of the license plate recognition system 1 of the present invention. The license plate recognition method includes a step 51 and a step 52. A step 53, a step 54, a step 55, a step 56, a step 57, and a step 58.

In step 51, the first processing module 24 transmits one of the images to be processed captured by the first image capturing module 23 to the server 3 through the first communication module 21.

In step 52, after receiving the to-be-processed image through the second communication module 31, the second processing module 33 obtains the to-be-processed using the first object detection model according to the to-be-processed image One character detection result corresponding to the image.

Referring to FIG. 4, it is worth noting that step 52 further includes a sub-step 521, a sub-step 522, a sub-step 523, a sub-step 524, and a sub-step 525.

In sub-step 521, the second processing module 33 uses the second object detection model according to the image to be processed to obtain a vehicle detection result corresponding to the image to be processed.

In sub-step 522, the second processing module 33 determines whether the vehicle detection result indicates that the to-be-processed image includes a to-be-processed vehicle block corresponding to a vehicle. When the second processing module 33 determines that the vehicle detection result indicates that the to-be-processed image includes the to-be-processed vehicle block, the process substep 523 is performed; when the second processing module 33 determines that the vehicle is to be detected When the measurement result indicates that the to-be-processed image does not include the to-be-processed vehicle block, the process returns to step 521 of the process. Wherein, the to-be-processed vehicle block is defined by a starting point coordinate of the vehicle in the to-be-processed image, a vehicle horizontal distance, and a vehicle vertical distance, but not limited thereto.

In sub-step 523, the second processing module 33 uses the third object detection model according to the vehicle block to be processed to obtain a license plate detection result corresponding to the vehicle block to be processed.

In sub-step 524, the second processing module 33 determines whether the license plate analysis result indicates that the vehicle block to be processed includes a block to be processed corresponding to a license plate. When the second processing module 33 determines that the license plate analysis result indicates that the to-be-processed vehicle block includes the to-be-processed license plate block, flow sub-step 525 is performed; when the second processing module 33 determines the license plate analysis When the result indicates that the to-be-processed vehicle block does not contain the to-be-processed license plate block, the process returns to step 521 of the process. Wherein, the pending license plate block is defined by a starting point coordinate of the license plate located in the pending vehicle block, a horizontal distance of a license plate and a vertical distance of a license plate, but not limited thereto.

In sub-step 525, the second processing module 33 uses the first object detection model according to the to-be-processed license plate block to obtain the character detection result corresponding to the to-be-processed license plate block.

It is worth noting that step 52 can also directly use the first object detection model to obtain the character detection result corresponding to the image to be processed according to the image to be processed. At this time, each first training data used to train the first object detection model includes the image and first position information indicating the block where each license plate character in the image is located.

In step 53, the second processing module 33 determines whether the character detection result indicates that the image to be processed includes a plurality of character block to be processed corresponding to a plurality of license plate characters of a license plate. When it is determined that the character detection result indicates that the to-be-processed image contains the waiting-for-processing character block, proceed to step 54; when it is determined that the character detection result indicates that the to-be-processed image does not contain the While waiting for the character block to be processed, return to step 52 of the process to determine the next image to be processed. Wherein, each character block to be processed is defined by a coordinate starting point of a character, a horizontal distance of a character, and a vertical distance of a character, but not limited to this.

），該第二處理模組33獲得該待處理字元區塊所對應的一代表點（如圖3所示之

）。其中，每一字元圖像區塊所對應的一代表點係為該字元圖像區塊所對應的一中心點。 In step 54, for each character block to be processed in the character detection result (as shown in Figure 3

), the second processing module 33 obtains a representative point corresponding to the character block to be processed (as shown in FIG. 3)

). Wherein, a representative point corresponding to each character image block is a central point corresponding to the character image block.

）。其中，該第二處理模組33係根據每一代表點，利用一最小平方法，獲得一回歸直線並作為該直線。 In step 55, the second processing module 33 obtains the straight line closest to each representative point (as shown in FIG. 3

). Wherein, the second processing module 33 uses a least squares method to obtain a regression line according to each representative point and uses it as the line.

），獲得一校正角度（如圖3所示之

）。其中，該參考線係為一水平線。 In step 56, the second processing module 33 according to the straight line and a preset reference line (as shown in Figure 3

) To obtain a correction angle (as shown in Figure 3

). Wherein, the reference line is a horizontal line.

In step 57, for each character block to be processed, the second processing module 33 corrects the character block to be processed according to the correction angle to obtain a corrected character block. Wherein, the second processing module 33 uses a rotation matrix to correct each character block to be processed according to the correction angle.

In step 58, for each corrected character block, the second processing module 33 uses a conventional image character recognition model to identify the corrected character block based on the corrected character block Corresponding characters of the license plate. In particular, when the servo terminal 3 includes a second display module electrically connected to the second processing module 33, each recognized license plate character is also displayed on the second display module.

It is worth noting that another embodiment of the license plate recognition method performed by the license plate recognition system 1 of the present invention is that the first processing module 24 (the electronic device 1) executes steps 51-58, and finally the first processing The module 24 transmits the license plate character corresponding to each correction character block identified in step 58 to the server 3 through the first communication module 21.

In summary, the new vehicle license plate recognition system 1 uses the first processing module 24 or the second processing module 33 to use the second object detection model and the third object detection model to obtain The to-be-processed license plate block in the to-be-processed image, and then using the first object detection model to obtain the character detection result corresponding to the to-be-processed license plate block, and indicating according to the character detection result Extract the straight line obtained from the representative point corresponding to each character block to be processed, and correct each character block to be processed into the correction character according to the correction angle defined by the straight line and the reference line Block, and finally identify the license plate character corresponding to each correction character block. Therefore, it can indeed achieve the purpose of new cost.

However, the above are only examples of the new model. When the scope of the new model cannot be limited by this, any simple equivalent changes and modifications made according to the patent application scope and patent specification content of the new model are still regarded as Within the scope of this new patent.

:待處理字元區塊

:代表點

:直線

:參考線

:校正角度 51~58:步驟 521~525:子步驟200: Communication network 1: License plate recognition system 2: Electronic device 21: First communication module 22: First storage module 23: First image capture module 24: First processing module 3: Server 31: No. Second communication module 32: second storage module 33: second processing module

: Pending character block

: Representative point

:straight line

:reference line

: Correction angle 51~58: Step 521~525: Substep

的代表點

、最接近每一代表點

的一直線

，以及該直線

與一參考線

所界定的一校正角度

；及 圖4是一流程圖，說明該車牌辨識方法如何獲得一字元偵測結果的細部流程。 Other features and functions of the present invention will be clearly presented in the embodiment with reference to the drawings, in which: FIG. 1 is a block diagram illustrating an embodiment of the novel license plate recognition system; FIG. 2 is a flowchart illustrating A license plate recognition method executed in this embodiment; FIG. 3 is a schematic diagram illustrating each character block to be processed in the license plate recognition method

Representative point

, Closest to each representative point

Line of

, And the line

With a reference line

A defined correction angle

; And Figure 4 is a flowchart illustrating the detailed process of how the license plate recognition method obtains a character detection result.

200: communication network

1: License plate recognition system

2: Electronic device

21: The first communication module

22: The first storage module

23: The first image capture module

24: The first processing module

3: server side

31: Second communication module

32: Second storage module

33: Second processing module

Claims (5)

A license plate recognition system, including: An image capturing unit, used to capture an image to be processed; A processing unit electrically connected to the image capturing unit; and According to the image to be processed, the processing unit uses a first object detection model for detecting a character block corresponding to each license plate character in the image to obtain the image to be processed Corresponding to a character detection result, the processing unit determines whether the character detection result indicates that the image to be processed includes a plurality of character block to be processed corresponding to a plurality of license plate characters of a license plate, when the processing When the unit determines that the character detection result indicates that the to-be-processed image includes the pending character block, for each pending character block in the character detection result, the processing unit obtains the pending Processing a representative point corresponding to the character block, the processing unit obtains a straight line closest to each representative point, the processing unit obtains a correction angle according to the straight line and a preset reference line, and for each word to be processed Meta block, the processing unit corrects the character block to be processed according to the correction angle to obtain a correction character block, and for each correction character block, the processing unit according to the correction character area Block, an image character recognition model is used to identify the license plate character corresponding to the corrected character block. The license plate recognition system according to claim 1, wherein a representative point corresponding to each character image block is a center point corresponding to the character image block, and the straight line is a regression straight line . The license plate recognition system according to claim 1, wherein the processing unit uses a second object detection model for detecting a vehicle block corresponding to a vehicle in the image based on the image to be processed To obtain a vehicle detection result corresponding to the to-be-processed image, the processing unit determines whether the vehicle detection result indicates that the to-be-processed image includes a to-be-processed vehicle block corresponding to a vehicle, when the processing unit When it is determined that the vehicle detection result indicates that the to-be-processed image has the to-be-processed vehicle block, the processing unit uses, according to the to-be-processed vehicle block, a corresponding one for detecting a license plate located in the image. The third object detection model of a license plate block obtains a license plate detection result corresponding to the pending vehicle block, and the processing unit determines whether the license plate analysis result indicates that the pending vehicle block includes a corresponding license plate Of a pending license plate block, when the processing unit determines that the license plate detection result indicates that the pending vehicle block includes the pending license plate block, the processing unit uses the An object detection model to obtain the character detection result corresponding to the license plate block to be processed. The license plate recognition system according to claim 1, wherein the processing unit includes a first processing module and a second processing module, the first processing module is electrically connected to the image capturing unit and to a communication network The first communication module, the second processing module is electrically connected to a second communication module connected to the communication network, the first processing module transmits the image to be processed captured by the image capturing unit to The second processing module, the second processing module uses the first object detection model according to the image to be processed to obtain the character detection result corresponding to the image to be processed, and the second processing module The group determines whether the image block to be processed contains the waiting character block, the second processing module obtains the representative point corresponding to each character block to be processed, and the second processing module obtains the straight line , The second processing module obtains the correction angle, the second processing module obtains the correction character blocks, and the second processing module recognizes the license plate characters corresponding to the correction character blocks . The license plate recognition system according to claim 1, wherein the processing unit includes a first processing module and a second processing module, the first processing module is electrically connected to the image capturing unit and to a communication network The first communication module, the second processing module is electrically connected to a second communication module connected to the communication network, the first processing module obtains the character detection result, and the first processing module determines the pending Whether the processed image contains the waiting character block, the first processing module obtains the representative point corresponding to each character block to be processed, the first processing module obtains the straight line, the first The processing module obtains the correction angle, the first processing module obtains the correction character blocks, and the first processing module recognizes and transmits the license plate characters corresponding to the correction character blocks to the first Two processing modules.

Images