Recogniform - Tecnologie di riconoscimento ottico


Go to the Italian website

MICR CMC7 Recognition Engine

 buy on-line     Recogniform Reader    download demo    download evaluation version     info  

Insieme di caratteri MICR CMC7 History: MICR CMC7 font was developed in France in late 1950s, for automatic cheque processing. It's currently used in Europe, Mexico, Brazil and in other countries world-wide.

Standard: digits' font print size is 2.85mm, while symbols' size is 2.70mm. Each character is made up by vertical lines divided by five narrow spaces and two larger spaces.

Capability: our MICR CMC7 recognition engine is size tolerant and it's able to read codelines with small size variation. The recognized character set is made up by the following digits and symbols:

Insieme di caratteri MICR CMC7

According to the picture order, the last symbols are called: S-1 (Internal), S-2 (Terminator), S-3 (Amount) , S-4 (Not Used), S-5 (Routing) and, after recognition, they are respectively reported like "/" "=" "!" ">" "^" in ASCII characters.

Accuracy: with MICR CMC7 recognition engine, the accuracy reaches 100%  in case of good quality scanned images; if the quality is low or medium, error rate will usually be less than 0.5%. For each recognized character a confidence level is reported, to let you to appreciate how  much accurate has been the recognition process.

Technology: our engine is based on three distinct classifiers using different methods: artificial neural network, features extraction, statistycal matrix matching. An expert system, usingfuzzy rules, reconciles all the outputs obtaining the best match.

Speed: speed is linked to cpu speed. Using Pentium processors, time performance are usually included between 50 and 150 CPS.

Input: monochrome images (like memory DIB handles, Device Independent Bitmap) at resolution of 200 or more DPI are accepted in input. Better results can be obtained using grayscaleimages with our dynamic thresholding SDK.

Output: output is made by recognized characters ASCII codes, by confidence level and by coordinates of each area containing characters.

Platform: all of the Windows 95/98/NT/Millennium/2000/Xp platforms are supported.

Packaging: the engine is packaged in DLL format. The total size is 500 KB. No extra file is required.

Here is a sample of CMC7 codeline image, automatically readable with 100% accuracy: 

Esempio di CMC7

To test this library you can download a demo application (500 KB) that allows you to recognize codelines from your images: please give us your feedback and your comments after the test!

Here is the source code sample that uses our engine:

// Demo user function
long RecognizeMyCodeline(long CodelineDIB)

  // Declare local variables 
  long Session,Error,Lines,LineLen;
  char Chars[1024]; // Codelines are less then 1024 chars !
  single Confidences[1024];
  rect Rects[1024]

  // Init the library in evaluation mode
  Error=CMC7_Init("demo","demo", &Session);
  // Check if any error occurred
  if (Error!=CMC7_OK)
    // Perform the recognition
    Error=CMC7_Read(Session, CodelineDIB,  TRUE);

    // Check if any error occurred
    if (Error!=CMC7_OK)

      // Retrieve the codeline length

      //You could allocate here memory for Chars, Confidences and Rects
      //using LineLen value instead to use fixed size buffers...

      // Retrieve the codeline chars

      // Retrieve the codeline char confidences...if required

      // Retrieve the codeline char rects...if required

      // =============================================
      // Add here your code to use the recognized data
      // =============================================


    // Deinitialize the library
  return (Error);

Pricing & Ordering Info 
You can buy this product in our
 on-line store.
To order this product, please contact us.

Evaluation Version
You can download an evaluation version of this product for
 Visual Basic, Visual C++ o Delphi

More Info
If you need further information about this product please use the
 contacts page.


2000-2022 Recogniform Technologies SpA - All rights reserved