IMG2TXT

A free opensource Android app to extract text from images using OCR technology.

• works offline (without Internet access)
• works on all versions of Android; from Android 6 up to Android 14 (the latest version release)
• get image from gallery or from Camera for extracting text
• crop image before OCR processing
• share image into IMG2TXT app to extract text from it
• app UI supports Arabic & English
• app supports extracting Arabic & English languages
• app supports extracting Arabic+English text from the same image
• Use three OCR engines (ML Kit -> Google Vision -> Tesseract OCR)
• color-coded confidence/accuracy extracted text
  • +80% -> white/black
  • 50-79% -> purple
  • -50% -> red
• let the user choose the text language before processing
• show the newlines in the result text
• ability to edit extracted text
• ability to copy extracted text

Install the img2txt app from Google Play: https://play.google.com/store/apps/details?id=com.softwarepharaoh.img2txt

documented fixes

Google Play : "This App Bundle contains native code, and you've not uploaded debug symbols. We recommend you upload a symbol file to make your crashes and ANRs easier to analyze and debug. Learn More". I documented the fix on my tech blog here.

Roadmap | timeline of release versions with tasks

• Old Codebase
  • use onActivityResult (modern code)
  • use general/universal openUrl() method/function
  • better way of cropping images
  • share image into IMG2TXT app to extract text from it
  • translations (English & Arabic), default is English
  • About Me
  • rate us on Google Play (removed)
  • link to Google Play
• v2.5.1
  • first opensource version release is v2.5.1 on April 21st, 2024
  • support Android 6 (Marshmallow) (SDK 23)
  • support Android 7
  • support Android 8
  • support Android 9
  • support Android 10
  • support Android 11
  • support Android 12
  • support Android 13
• v2.6.0
  • support Android 14 (SDK 34)
  • show thresholded/cleaned image (created by Tesseract)
• v2.7.0
  • release v2.7.0 on May 29th, 2024
  • show bounding rectangles/boxes around each recognized word
• v2.8.0
  • image to text OCR app release v2.8.0 on May 31st, 2024 with enhanced user experience and bug fixes.
  • simplify app UI layout
  • new simpler theme
  • show alert/notice if the mean_confidence of result text is less than 60%
  • on-device Tesseract OCR (English & Arabic models)
  • on-device Google vision API (latin scripted languages)
  • on-device ML Kit (latin scripted languages)
  • in case of Arabic language or both (Arabic+English), use Tesseract OCR
  • fallback strategy in case of English language is ML Kit -> Google Vision -> TesseractOCR
  • color-coded confidence/accuracy of the result text from ML Kit & Tesseract OCR
  • prompt the app user to choose the language of text on the image before processing it
  • customize crop-screen
  • make result/extracted text editable
  • add line breaks (newline) in extracted text (ML Kit & Tesseract OCR)
• v2.9.0
  • support Android 15 (SDK 35)
  • save OCR history (aka : Detailed scanned images history)
• v2.10.0
  • fix: make sure OCR history is saved in local db
  • feature: add reward ad (to collect points)
  • ui: add coins count
  • add 3 points every day
  • 1 image scan consumes 1 point
  • watch reward ads to gain points
  • save points in sharedPreferences
  • upgrade Tesseract4Android lib
• Next Version
  • rewrite app in Jetpack Compose
  • batch processing (in bulk)
  • PDF -> Images.foreach(ocr)
  • expose more functions to Java : cpp files in tesseract4android/src/main/cpp/tesseract and java files in tesseract4android/src/main/java/com/googlecode/tesseract/android must be modified/added.
  • support Hindi ( Indian language )
  • support Farsi ( Persian language )
  • save extracted text as PDF
  • choose more than one image to OCR
  • pre-process image with thresholding for more clarity and better results/accuracy/extracted text
  • Ability to edit image before/after running OCR on it (manual)

Resources & references

• Pre-Recognize Library - library with algorithms for improving OCR quality.
• using Tesseract in Python with OpenCV pre-processing

code snippets

public Bitmap convertGray(Bitmap bitmap3) {
        colorMatrix = new ColorMatrix();
        colorMatrix.setSaturation(0);
        ColorMatrixColorFilter filter = new ColorMatrixColorFilter(colorMatrix);

        Paint paint = new Paint();
        paint.setColorFilter(filter);
        Bitmap result = Bitmap.createBitmap(bitmap3.getWidth(), bitmap3.getHeight(), Bitmap.Config.ARGB_8888);
        Canvas canvas = new Canvas(result);

        canvas.drawBitmap(bitmap3, 0, 0, paint);
        return result;
    }

   /**
     * Binarization
     *
     * @param tmp Binarization threshold Default 100
     */
    private Bitmap binaryzation(Bitmap bitmap22, int tmp) {
        // Get the width and height of the image
        int width = bitmap22.getWidth();
        int height = bitmap22.getHeight();
        // Create a binary image
        Bitmap bitmap;
        bitmap = bitmap22.copy(Bitmap.Config.ARGB_8888, true);
        // Traverse the original image pixels and perform binarization processing
        for (int i = 0; i < width; i++) {
            for (int j = 0; j < height; j++) {
                // Get the current pixel value
                int pixel = bitmap.getPixel(i, j);
                // Get the value of the alpha channel
                int alpha = pixel & 0xFF000000;
                // Get the value of Red
                int red = (pixel & 0x00FF0000) >> 16;
                // Get the value of Green
                int green = (pixel & 0x0000FF00) >> 8;
                // Get the value of Blue
                int blue = pixel & 0x000000FF;

                if (red > tmp) {
                    red = 255;
                } else {
                    red = 0;
                }
                if (blue > tmp) {
                    blue = 255;
                } else {
                    blue = 0;
                }
                if (green > tmp) {
                    green = 255;
                } else {
                    green = 0;
                }

                // The optimal pixel value is calculated by weighted average algorithm.
                int gray = (int) ((float) red * 0.3 + (float) green * 0.59 + (float) blue * 0.11);
                // Set the image to black and white
                if (gray <= 95) {
                    gray = 0;
                } else {
                    gray = 255;
                }
                // Get the new pixel value
                int newPiexl = alpha | (gray << 16) | (gray << 8) | gray;
                // Assign pixels to the new image
                bitmap.setPixel(i, j, newPiexl);
            }
        }
        return bitmap;
    }

private Runnable runnable = new Runnable() {
        @Override
        public void run() {
            final Bitmap bitmap_1 = convertGray(BitmapFactory.decodeFile(path));

            baseApi.setImage(bitmap_1);
            result = baseApi.getUTF8Text();
            baseApi.recycle();

            handler.post(new Runnable() {
                @Override
                public void run() {
                    imageView2.setImageBitmap(bitmap_1);
                    textView.setText(result);
                    dialog.dismiss();
                }
            });
        }
    };