Optical mark recognition

To read an answer sheet, test, survey, election ballot, or other completed OMR form, digitize it in one of the supported formats. For best results, we recommend using a flatbed scanner (a basic office scanner or a copier will be enough). If you do not have a scanner, simply take a picture of the form with a smartphone camera.

If using a smartphone, hold it parallel to the paper and make sure the sheet is well lit with minimal glares and gradients.
The page must cover the entire area of the photo, with all 4 black squares in the corners present on the image.

The recognition is a 3-step process.

1. Initializing the recognition engine

Aspose.OMR for Java recognition engine is initialized with the recognition pattern (a file with .OMR extension), saved along with the printable form. The recognition pattern is loaded using GetTemplateProcessor method:

System::SharedPtr<Api::OmrEngine> engine = System::MakeObject<Api::OmrEngine>();
System::SharedPtr<Api::TemplateProcessor> processor = engine->GetTemplateProcessor(u"pattern.omr");

Recovering a recognition pattern file

If you have lost the recognition pattern file for the completed form, simply generate it again from the form’s source code using exactly the same paper size, font, and other layout setting.

2. Recognizing OMR forms

To recognize a form, process the scanned image or photo through RecognizeImage method of the initialized recognition engine:

System::SharedPtr<Api::OmrEngine> engine = System::MakeObject<Api::OmrEngine>();
System::SharedPtr<Api::TemplateProcessor> processor = engine->GetTemplateProcessor(u"pattern.omr");
System::SharedPtr<Model::RecognitionResult> result = processor->RecognizeImage(u"completed-form.jpg", 40);

The recognition method support accuracy adjustments for reliable results for different types of marks, even under difficult conditions (dirt, glare, and so on).

3. Saving recognition results

After the recognition is finished, you can get results in the most popular data storage formats.

Saving as CSV

Comma-separated values (CSV) is a lightweight text format that uses a comma to separate values in a table-like structure. Best suited for spreadsheet applications and simple relational database tables.

To save recognition results in CSV format, call GetCsv method of the recognition result object:

System::SharedPtr<Api::OmrEngine> engine = System::MakeObject<Api::OmrEngine>();
System::SharedPtr<Api::TemplateProcessor> processor = engine->GetTemplateProcessor(u"pattern.omr");
System::SharedPtr<Model::RecognitionResult> result = processor->RecognizeImage(u"completed-form.jpg", 40);
System::String resultCsv = result->GetCsv();

Saving as JSON

JSON is the most popular popular open standard notation for describing nested data structures. Best suited for NoSQL databases or web.

To save recognition results in JSON format, call GetJson method of the recognition result object:

System::SharedPtr<Api::OmrEngine> engine = System::MakeObject<Api::OmrEngine>();
System::SharedPtr<Api::TemplateProcessor> processor = engine->GetTemplateProcessor(u"pattern.omr");
System::SharedPtr<Model::RecognitionResult> result = processor->RecognizeImage(u"completed-form.jpg", 40);
System::String resultCsv = result->GetJson();

Generating the template Appendices