// Scan the photo for text using the tess-two API
public static String scanPhoto(Bitmap bitmap) {
TessBaseAPI baseApi = new TessBaseAPI();
baseApi.setDebug(true);
baseApi.init(DATA_PATH, lang);
// set the black list
baseApi.setVariable("tessedit_char_blacklist", "':;,.?/\\}][{!@#$%^&*()-_=+~");
baseApi.setVariable("save_blob_choices", "T");
baseApi.setImage(bitmap);
String recognizedText = baseApi.getUTF8Text();
// Iterate over the results and print confidence values for debugging purposes
final ResultIterator iterator = baseApi.getResultIterator();
String lastUTF8Text;
float lastConfidence;
iterator.begin();
do {
lastUTF8Text = iterator.getUTF8Text(PageIteratorLevel.RIL_WORD);
lastConfidence = iterator.confidence(PageIteratorLevel.RIL_WORD);
if (lastConfidence > 50) {
Log.d(TAG, String.format("%s => %.2f", lastUTF8Text, lastConfidence));
}
} while (iterator.next(PageIteratorLevel.RIL_WORD));
baseApi.end();
Log.d(TAG, recognizedText);
return recognizedText;
}
@SmallTest
public void testChoiceIterator() {
final String inputText = "hello";
final Bitmap bmp = TessBaseAPITest.getTextImage(inputText, 640, 480);
// Attempt to initialize the API.
final TessBaseAPI baseApi = new TessBaseAPI();
baseApi.init(TessBaseAPITest.TESSBASE_PATH, TessBaseAPITest.DEFAULT_LANGUAGE);
baseApi.setPageSegMode(TessBaseAPI.PageSegMode.PSM_SINGLE_LINE);
baseApi.setVariable(TessBaseAPI.VAR_SAVE_BLOB_CHOICES, TessBaseAPI.VAR_TRUE);
// Ensure that text is recognized.
baseApi.setImage(bmp);
String recognizedText = baseApi.getUTF8Text();
assertTrue("No recognized text found.", recognizedText != null && !recognizedText.equals(""));
// Iterate through the results.
ResultIterator iterator = baseApi.getResultIterator();
List<Pair<String, Double>> choicesAndConfidences;
iterator.begin();
do {
choicesAndConfidences = iterator.getChoicesAndConfidence(PageIteratorLevel.RIL_SYMBOL);
assertNotNull("Invalid result.", choicesAndConfidences);
for (Pair<String, Double> choiceAndConfidence : choicesAndConfidences) {
String choice = choiceAndConfidence.first;
Double conf = choiceAndConfidence.second;
assertTrue("No choice value found.", choice != null && !choice.equals(""));
assertTrue("Found an incorrect confidence value.", conf >= 0 && conf <= 100);
}
} while (iterator.next(PageIteratorLevel.RIL_SYMBOL));
iterator.delete();
assertNotNull("No ChoiceIterator values found.", choicesAndConfidences);
// Attempt to shut down the API.
baseApi.end();
bmp.recycle();
}
private void testGetUTF8Text(String language, int ocrEngineMode) {
final String inputText = "hello";
final Bitmap bmp = getTextImage(inputText, 640, 480);
// Attempt to initialize the API.
final TessBaseAPI baseApi = new TessBaseAPI();
baseApi.init(TESSBASE_PATH, language, ocrEngineMode);
baseApi.setPageSegMode(TessBaseAPI.PageSegMode.PSM_SINGLE_LINE);
baseApi.setImage(bmp);
// Ensure that the result is correct.
final String outputText = baseApi.getUTF8Text();
assertEquals("\"" + outputText + "\" != \"" + inputText + "\"", inputText, outputText);
// Ensure getRegions() works.
final Pixa regions = baseApi.getRegions();
assertEquals("Found incorrect number of regions.", regions.size(), 1);
regions.recycle();
// Ensure getTextlines() works.
final Pixa textlines = baseApi.getTextlines();
assertEquals("Found incorrect number of textlines.", textlines.size(), 1);
textlines.recycle();
// Ensure getStrips() works.
final Pixa strips = baseApi.getStrips();
assertEquals("Found incorrect number of strips.", strips.size(), 1);
strips.recycle();
// Ensure getWords() works.
final Pixa words = baseApi.getWords();
assertEquals("Found incorrect number of words.", words.size(), 1);
words.recycle();
// Ensure getConnectedComponents() works.
final Pixa connectedComponents = baseApi.getConnectedComponents();
assertTrue("Connected components not found.", connectedComponents.size() > 0);
connectedComponents.recycle();
// Iterate through the results.
final ResultIterator iterator = baseApi.getResultIterator();
String lastUTF8Text;
float lastConfidence;
int[] lastBoundingBox;
Rect lastBoundingRect;
int count = 0;
iterator.begin();
do {
lastUTF8Text = iterator.getUTF8Text(PageIteratorLevel.RIL_WORD);
lastConfidence = iterator.confidence(PageIteratorLevel.RIL_WORD);
lastBoundingBox = iterator.getBoundingBox(PageIteratorLevel.RIL_WORD);
lastBoundingRect = iterator.getBoundingRect(PageIteratorLevel.RIL_WORD);
count++;
} while (iterator.next(PageIteratorLevel.RIL_WORD));
iterator.delete();
assertEquals("Found incorrect number of results.", count, 1);
assertEquals("Found an incorrect result.", lastUTF8Text, outputText);
assertTrue("Result was not high-confidence.", lastConfidence > 80);
assertTrue("Result bounding box not found.", lastBoundingBox[2] > 0 && lastBoundingBox[3] > 0);
boolean validBoundingRect =
lastBoundingRect.left < lastBoundingRect.right
&& lastBoundingRect.top < lastBoundingRect.bottom;
assertTrue("Result bounding box Rect is incorrect.", validBoundingRect);
// Attempt to shut down the API.
baseApi.end();
bmp.recycle();
}
