Extract the car number from the image Version 4 (python)

Action group: Robin OCR

Description

The action extracts license plates from the image

Action icon

Parameters and their settings

PropertyDescriptionTypeFilling exampleMandatory field
Parameters
ContextAction context for working with Robin OCR 2.Robin.OCR2
Yes
File pathThe path to the file from which you want to extract the car number. The image must contain the machine number, otherwise the extracted data may be incorrect. Supported image formats: jpg, jpeg, bmp, dib, rle, jpe, jfif, tif, tiff, png.Robin.FilePath
Yes
Page numberThe page number of the file to extract data from. Numbering starts from 1. If the field is not filled in, the data will be extracted from the first page.Robin.Numeric
No
Folder pathThe path to the folder where the recognized image with the superimposed blocks into which the action divided the data will be saved. To save the file, you also need to fill in the «File name» field.Robin.FolderPathC:\doc\imgNo
File nameThe name of the image file with superimposed blocks (without extension). By default, a file with the extension *.png is created. To save the file, you also need to fill in the «Folder path» field.Robin.String
No
OverwriteIf the value is «true» and an image file with the same name and extension exists in the specified folder, then the new file will overwrite it. If «false», the file will not be overwritten and the action will return an error.Robin.BooleantrueNo
Results
The car numberThe car number extracted from the original image.Robin.String

Image pathThe path to the image file with superimposed blocks.Robin.FilePath

Recognition confidenceImage recognition accuracy in the range from 0 to 1.Robin.Numeric

Status

Status of license plate number recognition. 
Possible status variants:

  • Number fully recognized
  • Number partially recognized
  • Region not recognized
  • No license plate number found
  • Number recognized incorrectly because there are several numbers in the image

Number fully recognized - when the license plate number has been fully recognized.

Number partially recognized - when the number is not fully recognized: glare on the image, dirt on the number, part of the number is covered, digits are smudged, etc.

Region not recognized - the action returned the license plate number itself, but failed to retrieve only the license plate number region.

No license plate number found - the image has no text or the number contains invalid characters for a license plate number. e.g. any letters except A, B, E, K, M, H, O, P, C, T, Y, X. 

Number recognized incorrectly because there are several numbers in the image - the action found several license numbers, merged the data, the result is incorrect.

Robin.String

The decoding of the image recognition results by status is shown in the table below.

Results of image recognition by statuses


Status

Image quality

Recognition results

1Number fully recognizedAn image that clearly shows the license plate number, no other license plates, glare, etc.
  • License plate number - recognized license plate number,
  • Recognition accuracy - coefficient of obtained image accuracy,
  • Image with blocks - path to the image,
  • Status - Number fully recognized
2Number partially recognizedThere is glare on the image, dirt on the license plate, part of the license plate is covered, numbers are smudged, etc.
  • License plate number - recognized license plate number,
  • Recognition accuracy - coefficient of the obtained image accuracy, 
  • Image with blocks - path to the image,
  • Status - Number partially recognized
3Region not recognizedAn image in which only the region of the license plate cannot be recognized: it is covered, in mud, etc.
  • License plate number - recognized license plate number,
  • Recognition accuracy - coefficient of obtained image accuracy
  • Image with blocks - path to the image,
  • Status - Region not recognized
4No license plate number foundThe image has no text or the license plate number contains invalid characters for a license plate. e.g. any letters other than A, B, E, K, M, H, O, P, C, T, Y, X.
  • License plate number - recognized license plate number,
  • Recognition accuracy - coefficient of the obtained image accuracy, 
  • Image with blocks - path to the image,
  • Status - No license plate number found
5Number recognized incorrectly because there are several numbers in the imageThere are several objects in the image where you can recognize license plates and text. Signs, advertisements, other license plates and telephone numbers
  • License plate number - recognized license plate number,
  • Recognition accuracy - coefficient of obtained image accuracy
  • Image with blocks - path to the image,
  • Status - Number recognized incorrectly because there are several numbers in the image

Special conditions of use

  1. If an image without license plates or with several license plates is input, the text data from the image may be recognized incorrectly or not recognized at all. Therefore, it is recommended to use the action only for recognizing one license plate number.
  2. Acceptable image formats: jpg, jpeg, bmp, dib, rle, jpe, jfif, tif, tiff, png, pdf. If a file of a different extension is specified, the robot will terminate with an error. 
  3. If multiple numbers are found in the image, the action will find all numbers, merge the data, and return 1 invalid result.
  4. If the parameters "Folder path" and "File name" are not filled in, the file with overlapped blocks will not be saved.
  5. If "File name" and "File path" parameters are filled in, the image with blocks will be saved in the specified folder with the specified name in .png format.
  6. If the action does not find or recognize the license plate, the robot will terminate without error. The action will return "License plate number" - empty, "Recognition accuracy" - the coefficient of the obtained image accuracy, "Image with blocks" - the path to the image without blocks.
  7. It should be noted that the license plate number may contain similar symbols (1-I, 0-o, B-o-0, 9-0, etc.). In this case, the number is recognized, but the characters themselves may be recognized different from the original ones - "1" as "I", 0 (zero) as "O", and so on. The recognition accuracy in this case may decrease.

Example of use

Task

You need to extract the license plate number from the image 

M704PO198, 

presented as

Solution

Use the action "Create context with Robin OCR 2", "Extract the car number from the image".

Implementation

  1. Assemble a robot scheme consisting of actions:
  2. Fill in the "Create context with Robin OCR 2" parameters according to the customization standards for this action.
  3. Set the parameters for the "Extract the car number from the image" action. To fill the context, click on , then click on and select the context of the action "Create context with Robin OCR 2 1" from the dropdown list.
  4. Click on the "Start" button in the top panel.

Result

The program robot completed successfully.


Image with blocks:


Additional attachments

https://confluence.itbs.it.ru/x/9437Eg

Instructions for local connection

  1. Click "Start."
  2. Select "Settings".
  3. Select "System".
  4. Select "About".

  5. Select "Advanced system settings".

  6. In the window that opens, select "Environment variables".
  7. Click "Create".

  8. Enter:

    Name - ROBIN_PY_OCRENGINE_PATH

    Value - http://10.68.10.31:8080/api/v1/ocr/gettext/abt

    and click "ОК".

  9. Restart the computer.

  10. Connect to the vpn.

  11. You can start the robot.