PLAN FILLED-IN INFORMATION FROM BANKCHECKS BASED ON PRIOR - - PowerPoint PPT Presentation

AUTOMATIC EXTRACTION OF FILLED-IN INFORMATION FROM BANKCHECKS BASED ON PRIOR KNOWLEDGE ABOUT LAYOUT STRUCTURE

Alessandro L. Koerich - CEFET/PR Lee Luan Ling - UNICAMP

PLAN

• Automatic Bankcheck Processing
• Characteristics of Checks
• Bankcheck Modeling
• Background and Printed Info Elimination
• Experimental Results
• Conclusion

Automatic Bankcheck Processing

• Millions of handwritten and machine printed

bankchecks have to be processed every day

• 260 millions p/m
• Handwritten or Machine Printed Bankcheck

Processing

• Only the information encoded in the MICR line can be

handled automatically

• bank, agency, account, check, serial and verification

codes

• The filled-in information is manually handled

Two Main Topics of Research

• Information Extraction

– check identification through MICR line – based on prior knowledge about layout structure – database

• background patterns
• customer’s data
• Information Processing

– Handwriting Recognition

• digit amount, worded amount, payee’s name, date, city

– Signature Verification

Characteristics of Bankchecks

• Using knowledge about the basic structure
• f a document to process any document of

the same type.

• Brazilian Checks

– Complex Layout Structure – Standard Size – Standard Layout

Bankcheck Modeling

• The division in blocks is not sufficient
• We must consider the overlapping of

information

• We propose that a check can be divided in

three layers

• Background Pattern
• Printed Information
• Filled-in Information

Background and Printed Information Elimination

• The background pattern and the printed

information only disturbs the processing of the filled-in information

• Goal

– Eliminate the background pattern without degrade the filled-in parts

Background and Printed Information Elimination (cont.)

• Position Adjustment

– Skew – Vertical – Horizontal

• Background Elimination

– Subtraction

• IWB(x,y) = ICD(x,y) - ICB(x,y)

Extraction of Filled-in Information

• Information introduced by bank’s customers

– Digit amount, worded amount, payee’s name, city and date

• Relies on prior knowledge about bankcheck

layout structure

• Check identification through MICR line
• Standardized Layout --- Template

Baselines Elimination

• Compute horizontal projection profiles
• The points with high values of PPh indicate

the position of baselines

• For these positions convert back pixels to

with pixels

Printed Characters Elimination

• Printed characters strings appearing under

the baseline dedicated to signature

• Customer’s name, register identification number
• Generate a binary image which contains the

similar information

• Database
• Subtraction
• IWP(x,y) = IWB(x,y) - IGP(x,y)

Experimental Results

• Real Brazilian bankcheck images
• 200 dpi and 256 gray levels
• 100 real bankcheck images

– 100 Financial Institutions – 25 different writers

Conclusions

• Method for extracting the filled-in

information from bankchecks

• Extraction of different items of information
• digit amount, worded amount, payee’s name, date,

city and signature

• Method provide satisfactory results
• Post-processing to improve quality
• Automatic Bankcheck Recognition System