/// <summary>
/// Takes an image, and runs it through Tesseract, returning
/// the cleaned string it detected.
/// </summary>
/// <returns>The ocr.</returns>
/// <param name="image">Image.</param>
private async Task <string[]> ProcessOCR(UIImage image)
{
if (!_tesseractInitialised)
{
//tesseract.SetPageSegmentationMode(PageSegmentationMode.SingleWord);
_tesseractInitialised = await tesseract.Init("eng");
tesseract.SetVariable("tosp_min_sane_kn_sp", "10");
if (!string.IsNullOrEmpty(OCRWhiteList))
{
tesseract.SetWhitelist(OCRWhiteList);
}
}
bool success = await tesseract.SetImage(image.AsPNG().AsStream());
if (!isDisposing)
{
if (success)
{
var results = tesseract.Results(PageIteratorLevel.Word);
return(results?.Select(u => u.Text).ToArray());
}
}
return(null);
}
public async void Sample4JpgWithSetVariable()
{
await _api.Init("eng");
_api.SetVariable("tessedit_char_whitelist", "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ");
using (var stream = LoadSample("sample4.jpg")) {
var result = await _api.SetImage(stream);
Assert.IsTrue(result);
Assert.IsTrue(_api.Text.Contains("Good font for the OCR"));
}
}
public async void Sample4JpgWithSetVariable()
{
await _api.Init("eng");
_api.SetVariable("tessedit_char_whitelist", "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ");
using (var stream = LoadSample("sample4.jpg")) {
var result = await _api.SetImage(stream);
Assert.IsTrue(result);
Assert.AreEqual(
"Good font for the OCR\nDingufrfom n Me am\nhe mm mm m cm\n\nGood 60m size for ocn\n\n", _api.Text);
}
}
