OCR Framework

• Real time OCR

****************** Setup ******************

Android Studio Version 1.3.2

1. Import the OCRAumentia.aar file

File -> New Module -> Import .JAR / .AAR Package

2. Add the new Module (OCRAumentia) as a dependency:

Right click on Project Name -> Open Module Settings -> Dependencies -> + -> Module Dependency -> OCRAumentia -> OK

****************** Init the framework ******************

// Get singleton instance
ocrAumentia = OCRAumentia.getmInstance();

Use the build in camera to real time analyse the output frames:

// Get singleton instance
// Init Text Recognition engine
ocrAumentia.init(this, API_KEY, OCRAumentia.SCREEN_ORIENTATION_PORTRAIT, PreviewSizeWidth, PreviewSizeHeight, true, ImageFormat.NV21, frame, 400);

Or analyse single images from the resources, assets, URL, etc

// Get singleton instance
// Init Text Recognition engine
ocrAumentia.init(this, API_KEY, 0);

…

/**
 * Analyse image from assets folder and extract recognised text
 */
private void processImageFromAssets()
{
    AssetManager assetManager = getAssets();

    InputStream istr;

    Bitmap bitmap       = null;

    try
    {
        istr    = assetManager.open("pic1.jpg");
        bitmap  = BitmapFactory.decodeStream(istr);
        // Make sure the image is decoded in the ARGB_8888 format
        bitmap  = bitmap.copy(Bitmap.Config.ARGB_8888, true);
    }
    catch (IOException e)
    {
        Log.e(HELLO_TAG, e.getMessage());
    }

    if ( bitmap != null )
    {
        ocrAumentia.processFrame(bitmap, false);
    }
}

****************** Implement OnTextRecognition methods ******************

Get a list with the matched words and their confidence:

@Override
    public void matchedWords(Map<String, Integer> wordConfidenceMap)
    {
        Iterator<Map.Entry<String, Integer>> iterator = wordConfidenceMap.entrySet().iterator();

        while (iterator.hasNext())
        {
            Map.Entry<String,Integer> pairs = iterator.next();
            Integer value   =  pairs.getValue();
            String key      = pairs.getKey();

            Log.d(HELLO_TAG, key + "--->" + value);
        }
    }

For debug purpose you can also get the analysed frame with bounding boxes surrounding the matched words:

@Override
    public void imageResult(int[] outputData, int width, int height, boolean isRotated)
    {
        if (outputData != null)
        {
            if ( mCustomCameraView == null )
            {
                mCustomCameraView = new ImageView(this);

                frame.addView(mCustomCameraView, new ViewGroup.LayoutParams(width, height));
            }

            if ( outputBitmap != null )
            {
                outputBitmap.recycle();
                outputBitmap = null;
            }

            outputBitmap = Bitmap.createBitmap(width, height, Bitmap.Config.ARGB_8888);

            // Bitmap from output data
            outputBitmap.setPixels(outputData, 0, width, 0, 0, width, height);

            if (outputBitmap != null)
            {
                if ( !isRotated )
                {
                    Matrix matrix = new Matrix();

                    matrix.postRotate(90);

                    Bitmap rotatedBitmap = Bitmap.createBitmap(outputBitmap, 0, 0, outputBitmap.getWidth(), outputBitmap.getHeight(), matrix, true);

                    mCustomCameraView.setImageBitmap(rotatedBitmap);
                }
                else
                {
                    mCustomCameraView.setImageBitmap(outputBitmap);
                }
            }
        }
    }

Check the API doc for more info

****************** API ****************** [api.aumentia.com](http://api.aumentia.com/ocr_android/)
****************** Min Android Version ****************** 14
************************* OCR Framework version ************************* 0.5
****************** License ****************** [LICENSE](https://github.com/aumentia/OCR-Aumentia-Android/blob/master/LICENSE)
****************** Bugs ****************** [Issues & Requests](https://github.com/aumentia/OCR-Aumentia-Android/issues)