License plate identification method and system thereof

Patent metadata

Field	Value
Publication number	US-11443535-B2
Application number	US-201916980747-A
Country	US
Kind code	B2
Filing date	Jan 21, 2019
Priority date	Mar 14, 2018
Publication date	Sep 13, 2022
Grant date	Sep 13, 2022

Abstract

A license plate identification method is provided, including steps of: obtaining a to-be-processed image including all characters on a license plate; extracting several feature maps corresponding to character features of the to-be-processed image through a feature map extraction module; for each of the characters, extracting a block and a coordinate according to the feature maps through a character identification model based on a neural network; and obtaining a license plate identification result according to the respective blocks and the respective coordinates of the characters.

First claim

The invention claimed is:

1. A license plate identification method, comprising steps of:
receiving a raw image;
extracting a historical background image through a foreground and background subtraction module;
comparing the raw image with the historical background image to determine an amount of image change; and
determining whether the amount of image change is greater than a predetermined value,
when the amount of image change is greater than the predetermined value, generating a to-be-processed image comprising all of characters on a license plate;
extracting a plurality of feature maps comprising character features of the to-be-processed image through a feature map extraction module;
for each of the characters, extracting a block and a coordinate according to the feature maps through a character identification model based on a neural network; and
obtaining a license plate identification result according to the respective blocks and the respective coordinates of the characters.

2. The license plate identification method as claimed in claim 1 , further comprising steps of:
obtaining a vehicle front image or a vehicle rear image from the raw image through a vehicle front image capturing module or a vehicle rear image capturing module by using a first image feature and a first classifier; and
obtaining the to-be-processed image comprising all of the characters according to the vehicle front image or the vehicle rear image through a license plate character region detection model.

3. The license plate identification method as claimed in claim 2 , further comprising steps of:
obtaining at least one character block from the vehicle front image or the vehicle rear image through the license plate character region detection model by using a second image feature and a second classifier; and
determining a magnification according to the number of character blocks,
wherein the to-be-processed image is obtained based on the character block according to the magnification.

4. The license plate identification method as claimed in claim 1 , further comprising steps of:
receiving license plate identification results;
dividing the plurality of license plate identification results into at least two groups according to a license plate grouping rule;
voting for each sub-identification result in each of the groups;
when, for each group, there is one sub-identification result having a voting score that is higher than a threshold value, generating a final license plate identification result according the sub-identification results; and
updating the character identification model according to the license plate identification result or the final license plate identification result,
wherein the license plate grouping rule comprises a license plate naming grouping rule, an English character region and number character region grouping rule, a dash grouping rule, and a character relative position grouping rule.

5. The license plate identification method as claimed in claim 4 , further comprising steps of:
assigning a weight to each license plate identification result according to a time sequence of all the license plate identification results; and
when, for each group, a weighted sum of one sub-identification result is greater than the threshold value, generating the final license plate identification result according to the sub-identification results.

6. A license plate identification system, comprising:
an image capturing apparatus configured to capture at least one raw image; and
a processor configured to:
receive the raw image from the image capturing apparatus;
extract a historical background image through a foreground and background subtraction module;
compare the raw image with the historical background image to determine the amount of image change;
determine whether the amount of image change is greater than a predetermined value;
when the amount of image change is greater than the predetermined value, generate a to-be-processed image comprising all of characters on a license plate according to the raw image;
obtain a plurality of feature maps comprising character features of the to-be-processed image through a feature map extraction module;
for each of the characters, extract a block and a coordinate according to the feature maps through a character identification model based on a neural network; and
obtain a license plate identification result according to the respective blocks and the respective coordinates of the characters.

7. The license plate identification system as claimed in claim 6 , wherein the processor is further configured to:
obtain a vehicle front image or a vehicle rear image from the raw image through a vehicle front image capturing module or a vehicle rear image capturing module by using a first image feature and a first classifier; and
obtain the to-be-processed image including all of the characters according to the vehicle front image or the vehicle rear image through a license plate character region detection model.

8. The license plate identification system as claimed in claim 7 , wherein the processor is further configured to:
obtain at least one character block from the vehicle front image or the vehicle rear image through the license plate character region detection model by using a second image feature and a second classifier; and
determine a magnification according to the number of character blocks,
wherein the to-be-processed image is obtained based on the character block according to the magnification.

9. The license plate identification system as claimed in claim 6 , wherein the processor is further configured to:
receive a plurality of license plate identification results;
divide the plurality of license plate identification results into at least two groups according to a license plate grouping rule;
vote for each sub-identification result in each of the groups;
when, for each group, there is one sub-identification result having a voting score that is higher than a threshold value, generate a final license plate identification result according the sub-identification results; and
update the character identification model according to the license plate identification result or the final license plate identification result,
wherein the license plate grouping rule comprises a license plate naming grouping rule, an English character region and number character region grouping rule, a dash grouping rule, and a character relative position grouping rule.

10. The license plate identification system as claimed in claim 9 , wherein the processor is further configured to:
assign a weight to each license plate identification result according to a time sequence of all the license plate identification results; and
when, for each group, a weighted sum of one sub-identification result is greater than the threshold value, generate the final license plate identification result according to the sub-identification results.

Assignees

• Delta Electronics Inc

Inventors

• Chen Yu-Ta
• Liang feng-ming
• Jheng Jing-Hong

Classifications

• G06V10/267 Primary

  by performing operations on regions, e.g. growing, shrinking or watersheds · CPC title

• G06V30/153

  using recognition of characters or words · CPC title

• G06V30/1916

  Validation; Performance evaluation · CPC title

• G06V10/758

  Involving statistics of pixels or of feature values, e.g. histogram matching · CPC title

• G06V10/82

  using neural networks · CPC title