Analyzing your prompt, please hold on...
An error occurred while retrieving the results. Please refresh the page and try again.
Different types of exams and questionnaires assume different types of marks. Answer sheets usually require the bubbles to be completely filled with a dark pen or marker. Ballot boxes are typically marked with checkmarks or crosses. Surveys and questionnaires are sometimes filled with a pencil so that corrections can be made. In addition, paper quality, watermarks, dirt, and even lighting conditions when photographing a completed form can reduce recognition accuracy.
All Aspose.OMR for C++ recognition methods support recognitionThreshold
parameter that allows you to fine-tune form processing and produce results with near 100% accuracy. This parameter is called recognition accuracy threshold.
From a technical standpoint, it defines the fill percentage of the bubble from 0%
(empty bubble) to 100%
(completely filled bubble). Lower values (25
-40
) allow even the lightest marks (such as pencil checkmarks and even dots) to be recognized. However, they may cause dirt, paper defects, or watermarks to be identified as false positives. Higher values (70
or more) require a more solid fill and may cause pencil marks, checks and crosses to be ignored.
70
, unless you require the entire bubble to be completely filled with a dark pen or a marker.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"ballot123.jpg", 30);
System::String resultCsv = result->GetCsv();
Analyzing your prompt, please hold on...
An error occurred while retrieving the results. Please refresh the page and try again.