Read text Version 11 (Python)

Action group: Text recognition

Description

The action performs text recognition on the image and returns it as a result

Action icon

Документация ROBIN > Read text Tesseract OCR > 2024-10-18_09-38-14.png

Settings

Property	Description	Type	Filling example	Mandatory field
Parameters
Image	Path to image file. Supported image formats: (jpeg, jpg, bmp, png, tif, tiff)	Robin.Image	C:\doc\img.png	Yes
Expected languages of text in the image	Expected languages of text in the image Expected languages of the text in the PDF file A dropdown list of items: Russian English Vietnamese Arabic Spanish Portuguese Indonesian Persian Turkish Kazakh Belarusian Default value - Russian	Robin.String		Yes
Additional language	An additional language required for document recognition A dropdown list of items: No Russian English Vietnamese Arabic Spanish Portuguese Indonesian Persian Turkish Kazakh Belarusian The default value is No If the same option is selected in the "Language" and "Additional Language" parameters, there will be no error. The duplicate will be counted as 1 language	Robin.Collection		No
Content format	Expected text content format. Available text formats: (Line, Block, Page)	Robin.String		Yes
Trained model	Tesseract trained model file in .taineddata format. Allows you to load your own model trained on the required fonts. If the parameter is populated, it will be prioritized over the "Language" and "Additional language" parameters
Options	Configuration options for OCR)	Robin.String		No
Results
Result	Text (string) recognized from image	Robin.String

Special conditions of use

The default mode in the "Parameters" field is --psm 3.

All parameters are listed with a space in the format --parameter value_parameter.

List of all parameters: https://muthu.co/all-tesseract-ocr-options/.

Parameter	Default value	Description
Main parameters
oem	3	0: Use only the outdated tesseract engine 1: Use only the LSTM neural network 2: Use the old tesseract engine and the LSTM neural network at the same time 3: Default value, what is available is used
psm	3	0: Orientation and Scenario Detection Only (OSD) 1: Automatic page segmentation using the OSD menu 2: Automatic page segmentation, but no OSD or OCR 3: Fully automatic page segmentation, but no on-screen menus 4: Assume that the text has one column of variable sizes 5: Assume that the text is a single homogeneous block of vertically aligned text 6: The image is treated as one single block of text 7: The image is treated as a single text string 8: The image is treated as a single word 9: The image is treated as one word in a circle 10: The image is treated as a single symbol 11: Sparse text. Search for as much text as possible in random order 12: Sparse text with on-screen menus 13: Unprocessed string. The image is treated as a single text string, bypassing Tesseract-specific hacks
Additional parameters
edges_min_nonhole	14	Minimum number of box pixels to recognize
textord_space_size_is_variable	0	If true (1) is set, word delimiter spaces are assumed to be of variable width, even if the characters are of fixed pitch
textord_tabfind_find_tables	1	Launch table detection
textord_force_make_prop_words	0	Apply proportional word segmentation to all strings
textord_width_limit	8	Maximum width of blocks for creating rows
tessedit_pageseg_mode	6	0: Orientation and scenario detection only 1: Auto mode + orientation detection 2: Auto mode 3: Column detection 4: Block detection 5: String detection 6: Word detection 7: Symbol detection
textord_max_noise_size	7	Maximum noise size in pixels
tessedit_dont_blkrej_good_wds	0	If true (1) is set, the word segmentation quality score is used
tessedit_char_blacklist		Blacklisting characters that cannot be recognized
tessedit_char_whitelist		White list of characters to recognize
List of chars to override tessedit_char_blacklist		List of symbols to override tessedit_char_blacklist

Example of use

Task

Read the text in the image

Solution

Use the "Read text" action

Implementation

Transfer the "Read text" action to the workspace.
Set the parameters of the "Read text" action
1. "Image" parameter. Specify the path to the image file, the text of which will be recognized.
2. Parameter "Expected languages of text in the image". Select the expected languages of the text in the image. The following languages are available for selection: "Russian language", "English language", "Russian and English language", "Spanish language", "Portuguese language". In this case, it is "Russian language".
3. "Content format" parameter. Select the expected text format. The following are available for selection: "Line", "Block", "Page". In this case - "Line".
4. "Result". The text obtained during text recognition from the image. Write the result to the "Text" variable.

2. Click on the "Start" button in the top panel.

Result

The program robot completed successfully. The text is read from the image.