| 摘要 |
PROBLEM TO BE SOLVED: To reduce decision errors due to the fluctuation of detection data which is caused by the secular change and partial stain of a bill and to discriminate the authenticity of paper sheet with high accuracy by using standard average and standard deviation which are preliminarily prepared at each position and evaluating statistically compensation sample data. SOLUTION: Data which extracts a surface pattern of a print design on the surface of sheet paper is acquired by converting detection data of sensors P0 to P4 into relative value data. Compensation sample data is acquired by normalizing the relative value data. Normalizing relative value data in this way enables appropriate data normalization processing which considers the indivisual error of the sensors P0 to P4 to be used and the individual difference of the paper sheet to be discriminated. Compensation sample data at each position is statistically evaluated by using standard average and standard deviation at each position which are preliminarily prepared. Specially, in different denominations like American dollar bills, the bills whose print design patterns look like one another can effectively be discriminated. |
