How to increase plain text processing speed


I want to maximize text recognition speed. I only need recognized text, extraction coordinates, the confidence rating for words and line segmentation, and I don't need any of the data on font, color, blocks or any of the other data. How can I do it?


Here are some basic tips on speeding up processing in FineReader Engine 12:

  • Call the LoadModule method with appropriate Engine module right after you create the Engine object. This prepares all the modules for processing and reduces the time it takes to process the first image.
  • Call FREngine.LoadPredefinedProfile("TextExtraction_Speed") before processing. More information in Help, chapter "API Reference > Engine Object (IEngine Interface) > Supplementary Methods > LoadPredefinedProfile Method of the Engine Object".
    Here is the list of settings, which are applied by TextExtraction_Speed profile:
    • [ObjectsExtractionParams]
      DetectPorousText = FALSE
      DetectTextOnPictures = TRUE
      EnableAggressiveTextExtraction = TRUE
      FastObjectsExtraction = TRUE
      ProhibitColorImage = TRUE
      RemoveGarbage = TRUE
      RemoveTexture = FALSE

    • [PageAnalysisParams]
      DetectPictures = FALSE
      EnableTextExtractionMode = TRUE
      ProhibitModelAnalysis = TRUE

    • [PrepareImageMode]
      CorrectSkew = FALSE
      DiscardColorImage = TRUE
      UseFastBinarization = TRUE

    • [RecognizerParams]
      FastMode = TRUE

    • [SynthesisParamsForDocument]
      DetectDocumentStructure = FALSE
      DetectFontFormatting = FALSE

    • [SynthesisParamsForPage]
      DetectFontFormattingAtPageLevel = TRUE
      DetectTextColor = TSPV_Yes
      DetectMatrixPrinter = FALSE

  • Make sure that input images come in recommended quality (300dpi, grayscale for letter-based languages and 400-600 dpi for hieroglyphic language). Low input quality slows down the processing.

Have more questions? Submit a request



Please sign in to leave a comment.