Android Example OCR OFFLINE - Tesseract

joilts · Aug 28, 2015

I'm working in a project that needs OCR offline. Made a small progress and decided to share here and get some feedback.
I've searched a little bit at this forum and google about it. Found options to use online OCR (NJDUDE's Lib or Erel's example). In the same project I also needs to manipulate some images and got DrewG Exemple about inline code to use JAVACV/OPENCV. This was the point to test Tesseract OCR in the same way (Inline code).
Downloaded the Lib at this link https://repo1.maven.org/maven2/org/bytedeco/javacpp-presets/1.0/javacpp-presets-1.0-bin.zip
More details about this can be found here.

Unzipped and copied the files I needed to my Additional Lib Folder.
The files I used: javacpp.jar, tesseract-android-arm.jar, leptonica-android-arm.jar, tesseract.jar, leptonica.jar

Coded Basic Example from bytedeco page. Made some changes to send the image as a file path to the image saved somewhere in the phone and got the "translation" text.

OBS: I needed to download tessdata files to my cell. Tried to add them to my app, but they were too big and I got some error deploying my app to cell (need to see this more carefully). The files to many languages can be found here or at google project page. I have download this one for my test example.

Here is the code I used. My test phone is a S4.

Hope it helps.

joilts · Dec 12, 2016

roberto64 said:
I thank you for your disponibiltà and the time you're losing me, I tried your routines is fine to do the image B & W, realizing edesso comene works that I can put inside other commands in rurine fact I added "IplConvKernel mat = cvCreateStructuringElementEx (5, 5, 2, 2, CV_SHAPE_RECT);
cvErode (grayImg, grayImg, mat, 1); so I more black letters and numbers, but I noticed that tessdata OCR does not recognize well the numbers and sometimes letters.
I thank you you're great.

Now you have to train your Tesseract and create your on font file to use at your app.

roberto64 · Dec 12, 2016

joilts said:
Now you have to train your Tesseract and create your on font file to use at your app.

hi, I used VietOCR.NET to understand what kind of fomato Tesseract OCR could recognize both letters and numbers, I made a crop the image and tried to make it recognize OCR, sarebe this what you say? or you must also make a Piccalo rutin like this (var = srcImage1 System.Drawing.Image.FromFile (@ "D: \ Image \ font_english.jpg");
var newWidth1 = (int) (srcImage1.Width * 2);
var newHeight1 = (int) (srcImage1.Height * 2);

var image = new Bitmap (srcImage1, new Size (newWidth1, newHeight1));
Tesseract OCR var = new ();

ocr.Init (@ "D: \ OCRTEST \ tessdata \", "eng", false);
ocr.SetVariable ("tessedit_char_whitelist", "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz-0123456789 '?.; =, ()");
var result = ocr.DoOCR (image, Rectangle.Empty);
foreach (Word word in result)
{
Response.Write (word.Text + "");

})
Greetings

DonManfred · Dec 12, 2016

roberto64 said:
var newWidth1

Please use CODE TAGS when posting code!

roberto64 · Sep 7, 2017

Hi, I do not understand why passing a Bitmap image within the function as described below, by mistake, I would like to go directly to the Bitmap image and not save the image first and then view it.
regards

B4X:

Dim path As Bitmap
        Dim retStr As String = ""
        retStr = NativeMe.RunMethod("getText",  Array (path,File.DirRootExternal))

#AdditionalJar: javacpp-presets-bin\javacpp
#AdditionalJar: javacpp-presets-bin\tesseract-android-arm
#AdditionalJar: javacpp-presets-bin\leptonica-android-arm
#AdditionalJar: javacpp-presets-bin\tesseract
#AdditionalJar: javacpp-presets-bin\leptonica

'public static String getText(String path, String filename, String extension, String TrainFileDir)
#If JAVA
import org.bytedeco.javacpp.*;
import static org.bytedeco.javacpp.lept.*;
import static org.bytedeco.javacpp.tesseract.*;
import android.graphics.Bitmap;
import android.graphics.BitmapFactory;
import android.graphics.Matrix;


private TessBaseAPI tessBaseAPI;
//public static String getText(String path, String filename, String extension, String TrainFileDir)
public static String getText(Bitmap path, String TrainFileDir) {
    BA.Log("" + "Here - getText() ");
    BytePointer outText;
    TessBaseAPI api = new TessBaseAPI();
    BA.Log("" + "Before Init ");
    int retCode = api.Init(TrainFileDir, "eng");
    BA.Log("directori =" + TrainFileDir);
    BA.Log("RETCODE =" + retCode);
    if (retCode != 0) {
        return("Could not initialize tesseract.");
    }
  
    PIX image = pixRead(Bitmap);
    BA.Log("" + "File Open");
    api.SetImage(image);
    BA.Log("" + "Before get Text");
    outText = api.GetUTF8Text();
    api.End();
    outText.deallocate();
    pixDestroy(image);
    return(outText.getString());

}
#End If

Android Example OCR OFFLINE - Tesseract

Attachments

joilts

Member

roberto64

Active Member

DonManfred

Expert

roberto64

Active Member

Similar Threads

Android Example OCR OFFLINE - Tesseract

Attachments

joilts

Member

roberto64

Active Member

DonManfred

Expert

roberto64

Active Member

Similar Threads

Privacy & Transparency

Privacy & Transparency