Method for fraud detection using multiple scan technologies

Abstract

A method is disclosed for detecting fraud. The method optically recognizes a character imprinted with a magnetic ink on a medium using a first optical character recognition algorithm. The method also optically recognizes the character using a second optical character recognition algorithm. In addition, the method magnetically recognizes the character using a magnetic recognition algorithm. The method determines if the character is potentially fraudulent based on the recognition results of the scan using the first optical character recognition algorithm, the scan using the second optical character recognition algorithm, and the magnetic scan. If the character is potentially fraudulent, the method communicates a fraud indicator.

1 . A method for detecting fraud, the method comprising:
 magnetically scanning a single row of Magnetic Ink Character Recognition (“MICR”) characters imprinted on a medium using magnetic ink, recognizing individual MICR characters within the single row of MICR characters, and detecting a magnetic signal strength of the magnetic ink of the MICR characters;   scanning the single row of MICR characters and recognizing individual MICR characters within the single row of MICR characters using a first Optical Character Recognition (“OCR”) algorithm;   scanning the single row of MICR characters and recognizing individual MICR characters within the single row of MICR characters using a second OCR algorithm, wherein the second OCR algorithm differs from the first OCR algorithm;   determining if a MICR character is potentially fraudulent based on the recognition results of the magnetic scan, the scan using the first OCR algorithm, and the scan using the second OCR algorithm, wherein determining that a MICR character is potentially fraudulent comprises one or more of determining that the magnetic signal strength of the MICR character is lower than a predetermined value and the character recognized by the magnetic scan, the character recognized by the scan using the first OCR algorithm, and the character recognized by the scan using the second OCR algorithm are not identical; and   communicating a fraud indicator in response to determining that a MICR character is potentially fraudulent.   
   
   
       2 . The method of  claim 1 , further comprising transporting the medium relative to a location where the medium is optically and magnetically scanned. 
   
   
       3 . The method of  claim 2 , further comprising re-transporting the medium based on the recognition results of the scan using the first OCR algorithm, the scan using the second OCR algorithm, and the magnetic scan. 
   
   
       4 . The apparatus of  claim 1 , wherein the medium is configured as a check. 
   
   
       5 . The apparatus of  claim 1 , wherein determining if a MICR character is potentially fraudulent further comprises using a look-up table. 
   
   
       6 . The apparatus of  claim 1 , wherein determining if a MICR character is potentially fraudulent further comprises using a decision tree. 
   
   
       7 . The apparatus of  claim 1 , wherein the recognition results of the magnetic scan, the scan using the first OCR algorithm, and the scan using the second OCR algorithm each comprise a plurality of character candidates each with a confidence level.