OCR convert jpg, png or TIFF output to text is unreadable

Hi,

I have a full (but temporary licence). My customer want to know if you can convert tax slip from a image into text. Ia have try so many time but the result is alwas unreadable.
I use Aspose.OCR v24.9.0

Here is the code I use :
Aspose.OCR.License license = new Aspose.OCR.License();
license.SetLicense(“D:\Code\Outil_OCR\Aspose.OCR-for-.NET-master\Application_Aspose_Ocr_Net_6\Aspose.OCR.Net\bin\Debug\net6.0\win-x64\Aspose.OCRfor.NET.lic”);

// Initialize an instance of AsposeOcr
AsposeOcr api = new AsposeOcr();

// Image Path
fullPath = @“D:\Code\AI_fiscal\CasTest\Exemple\T3.png”;

// Create OcrInput object and add document
OcrInput input = new OcrInput(InputType.SingleImage);
input.Add(fullPath, startPage: 0, pagesCount: 1);

// Recognize image
List results = api.Recognize(input, new RecognitionSettings { });

// Print result
int pageCounter = 0;
foreach (var result in results)
{
PrintRecognitionResult(result, pageCounter++);
}

Can you help please.

@Solariss73

It seems that you are experiencing issues with the readability of the text extracted from images using Aspose.OCR. Here are some suggestions to improve the OCR results:

  1. Preprocessing Filters: Consider applying preprocessing filters to enhance the image quality before recognition. This can include rotating, skew correction, or noise reduction. For example, you can use a PreprocessingFilter to rotate the image slightly if it is skewed.

    PreprocessingFilter filters = new PreprocessingFilter();
    filters.Add(PreprocessingFilter.Rotate(0.5f)); // Adjust the angle as needed
    
  2. Recognition Settings: Make sure to configure the recognition settings appropriately. You can specify the language of the text in the image, which can significantly improve accuracy. For instance, if the text is in English, set the language accordingly:

    RecognitionSettings settings = new RecognitionSettings();
    settings.Language = Language.Eng; // Set the appropriate language
    
  3. Image Quality: Ensure that the images you are using are of good quality. Low-resolution images or those with poor contrast can lead to unreadable text. If possible, use high-resolution scans or clear photographs.

  4. Testing with Different Images: If the current image is not yielding good results, try testing with different images to see if the issue persists. This can help determine if the problem is with the specific image or the OCR setup.

  5. Reviewing Documentation: For more detailed guidance on using Aspose.OCR effectively, you can refer to the official documentation, which provides comprehensive information on various features and best practices.

By implementing these suggestions, you should be able to improve the readability of the text extracted from your images. If you continue to face issues, consider sharing specific examples of the images and the results you are getting for further assistance.

Sources:
[1]: Get Recognition Result in OCR Image Recognition | Aspose.OCR .NET API
[2]: Aspose.OCR for .NET - Aspose Documentation

Didn’t help :frowning:

@Solariss73

Can we get a sample image for the tax slips that you want to process using the API? We will investigate at our end and share our feedback with you.