How to Call the tesseract mex function from MATLAB?

Here, we have explained the tesseract mex and MATLAB OCR (tesseract) function. Compared the both results with different images. We have compared the performance of the both OCR engine.

Tesseract mex

It has three input variables. The output of the mex variable only has the converted text. This mex was built by tesseract 3.01.

Below link contains the tesseract dependency files, unzip and place the file in C drive. Include the folder path in the system path.

Tesseract dependency Files

Please click the below link. It contains tesseract 32bit and 64bit.

Tesseract MEX File

1)      Image- Input image should be in 2-dimension uint8 format.

2)      Language- eng,jpn…etc

3)      Page Segmentation Mode

 

0 = Orientation and script detection (OSD) only.

1 = Automatic page segmentation with OSD.

2 = Automatic page segmentation, but no OSD, or OCR

3 = Fully automatic page segmentation, but no OSD. (Default)

4 = Assume a single column of text of variable sizes.

5 = Assume a single uniform block of vertically aligned text.

6 = Assume a single uniform block of text.

7 = Treat the image as a single text line.

8 = Treat the image as a single word.

9 = Treat the image as a single word in a circle.

10 = Treat the image as a single character.

MATLAB OCR

 Since 2014, this function was available in MATLAB. They have used Tesseract 3.02 version in OCR module. It has five inputs variables. The output of this function provides Text, Character level coordinates, Character level confidence score, Word Text, Word level coordinate and Word level Confidence Score.

1)      Image

2)      ‘Language’

3)      ‘English’ (or) other languages

4)      ‘TextLayout’

5)      ‘Auto’ or ‘Block’ or ‘Line’ or ‘Word’

 

Text Layout Types

Auto	Automatic Page Segmentation
Block	Assume a Single Uniform Block of Text
Line	Treat the image as a Single Text Line
Word	Treat the image as a Single Word

Please refer the below link for a brief explanation.

http://in.mathworks.com/help/vision/ref/ocr.html?searchHighlight=ocr&s_tid=doc_srchtitle#namevaluepairarguments

 

MATLAB Code

 

%% reading input image %%

rawImage=imread(‘sampleImage.jpg’);

%% converting the input image into grayscale image %%

if size(rawImage,3)==3

grayImage=rgb2gray(rawImage);

else

grayImage=rawImage

end

%% tesseract function calling %%

Text=tesseract(grayImage’,’eng’,3);

%%In build OCR function %%

Text0=ocr(grayImage,’Language’,’English’,’TextLayout’,’Auto’);

Performance Analysis

The document was scanned at 200 dpi, it has been used for Based on our analysis tesseract mex takes less time for OCR ( 216 DPI Document ). But it is not providing the confidence score and coordinates of the texts. Tesseract accuracy was lesser than the MATLAB OCR.

	Tesseract Mex	MATLAB OCR
Extraction Time	3.798140 seconds.	4.210153 seconds.
Character Level Confidence Score	No	Yes
Character Coordinates	No	Yes
Word Level Confidence Score	No	Yes
Word Coordinates	No	Yes

Input Sample used for testing:

Please find the below tesseract mex output file.

Tesseract MEX file output

Please find the below MATLAB OCR output file.

MATLAB OCR output