Tenprint card selector with fingerprint pattern level processing used to select candidates for sub-pattern level processing

• US 6,002,785 A
• Filed: 10/16/1996
• Issued: 12/14/1999
• Est. Priority Date: 10/16/1995
• Status: Expired due to Term

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Ucida, Kaoru
• Primary Examiner(s): Kelley, Christopher S.
• Assistant Examiner(s): CHAWAN, SHEELA C
• Application Number:
  US08/732,875
• Time in Patent Office: 1,154 Days
• Field of Search: 382/124, 382/125, 382/116, 382/126, 382/115, 340/825.34
• US Class Current: 382/124
• CPC Class Codes: G06V 40/1365 Matching; Classification

Abstract

A tenprint selector comprises a fingerprint pattern level feature difference calculating unit which receives and compares a set of candidates for pattern level feature types of fingerprint images shown on file cards and a set of candidates for pattern level feature types of fingerprint images shown on a search card and calculates a pattern level feature difference, a card difference calculating unit which calculates a card difference based on the pattern level feature difference of the fingerprint images of all fingers shown on the file cards and the search card, and a judging unit which compares the card difference determined by the card difference calculating means with a predetermined card difference threshold value and judges, whether the file card shall be selected to be further checked with the search card.

Claims

1. A tenprint card selector for selecting tenprint cards, which have features similar to those of a sample, from a file having a plurality of tenprint cards, comprising:

a fingerprint pattern level feature difference calculating means for receiving and comparing a set of candidates for pattern level feature types of fingerprint images shown on file cards and a set of candidates for pattern level feature types of fingerprint images shown on a search card and calculates a pattern level feature difference which shows the degree of similarity between the sets of candidates;

a card difference calculating means for calculating a card difference showing similarity of the features of the fingerprint images shown on said file cards and said search card based on said pattern level feature difference of the fingerprint images of all fingers shown on said file cards and said search card; and

a judging means for comparing the card difference determined by said card difference calculating means with a predetermined card difference threshold value and judging whether said file card shall be selected to be further checked with said search card,

wherein when the comparison of the sets of candidates for the pattern level feature types of said file cards and said search card results in that both sets of candidates are similar or at least either of the pattern level feature types is unclassifiable, that the sets of candidates do not have any common pattern level feature type, or that a plurality of candidates in the sets of candidates have the same pattern level feature types, said fingerprint pattern level feature difference calculating means determines the value of the pattern level feature difference to respectively be a minimum value, a maximum value, and a value appropriately determined between said minimum and maximum values, and

wherein said card difference calculating means determines a sum of all the pattern level feature differences of the respective fingers calculated by said fingerprint pattern level feature difference calculating means as the card difference.

2. A tenprint card selector for selecting tenprint cards, which have features similar to those of a sample, from a file having a plurality of tenprint cards, comprising:

a fingerprint pattern level feature difference calculating means for receiving and comparing pattern level feature types of fingerprint images shown on file cards and pattern level feature types of fingerprint images shown on a search card and calculates a pattern level feature difference to show whether or not said pattern level feature types are identical mutually;

a fingerprint sub-pattern level feature calculating means for receiving sub-pattern level information on the fingerprint images shown on said file cards and sub-pattern level information on the fingerprint images shown on said search card and calculating a sub-pattern level feature showing secondary features of said fingerprint images;

a fingerprint sub-pattern level feature difference calculating means for receiving and comparing the sub-pattern level feature of the file cards and the sub-pattern level features of the search card determined by said sub-pattern level feature calculating means and also receiving and referring to the pattern level feature types of the fingerprint images shown on said file cards and said search card, and calculating a sub-pattern level feature difference which shows the degree of similarity between said sub-pattern level features;

a card difference calculating means for calculating a card difference showing similarity of the features of the fingerprint images shown on said file cards and said search card based on said pattern level feature difference of the fingerprint images of all fingers shown on said file cards and said search card and said sub-pattern level feature difference of the fingerprint images of all fingers shown on said file cards and said search card; and

a judging means for comparing the card difference determined by said card difference calculating means with a predetermined card difference threshold value and judging whether said file card shall be selected to be further checked with said search card.

3. The tenprint card selector as set forth in claim 2, wherein

said fingerprint pattern level feature difference calculating means is provided in multiple numbers to individually relate to each finger and calculates a pattern level feature difference with respect to the fingerprint image of each related finger,

said fingerprint sub-pattern level feature calculating means is provided in multiple numbers to individually relate to each finger and calculates a sub-pattern level feature with respect to the fingerprint image of each related finger, and

said fingerprint sub-pattern level feature difference calculating means is provided in multiple numbers to individually relate to each finger and calculates a sub-pattern level feature difference with respect to the fingerprint image of each related finger.

4. The tenprint card selector as set forth in claim 2, wherein

when the comparison of the pattern level feature types of said file cards and said search card results in that both pattern level feature types are same or at least either of the pattern level feature types is unclassifiable, or that the pattern level feature types do not match, said fingerprint pattern level feature difference calculating means determines the value of the pattern level feature difference to be a minimum value, or a maximum value,

said fingerprint sub-pattern level feature calculating means receives said sub-pattern level information containing position coordinates of the singular points in the fingerprint images on said file cards and said search card to calculate a distance between said singular points and determines it as the sub-pattern level feature,

when the pattern level feature types of the fingerprint images on said file cards and the pattern level feature types of the fingerprint images on said search card are identical, said fingerprint sub-pattern level feature difference calculating means calculates said sub-pattern level feature difference based on the positional difference of the singular points which are sub-pattern level features according to a calculating formula determined according to said pattern level feature types, and

said card difference calculating means determines a sum obtained by adding all the pattern level feature differences of the respective fingers calculated by said fingerprint pattern level feature difference calculating means and the sub-pattern level feature differences of the respective fingers obtained by said fingerprint sub-pattern level feature difference calculating means as the card difference.

5. The tenprint card selector as set forth in claim 2, wherein

when the comparison of the pattern level feature types of said file cards and said search card results in that both pattern level feature types are same or at least either of the pattern level feature types is unclassifiable, or that the pattern level feature types do not match, said fingerprint pattern level feature difference calculating means determines the value of the pattern level feature difference to be a minimum value, or a maximum value,

said fingerprint sub-pattern level feature calculating means receives said sub-pattern level information containing position coordinates of the singular points on the fingerprint images of said file cards and said search card to calculate a distance between said singular points and determines it as the sub-pattern level feature,

when the pattern level feature types of the fingerprint images on said file cards and the pattern level feature types of the fingerprint images on said search card are loops and identical, whorls and identical, or others, said fingerprint sub-pattern level feature difference calculating means determines (the absolute value of a difference between the core-to-delta distance of the file card and the core-to-delta distance of the search card--prescribed value), (average value of the absolute values of differences between six core-to-delta distances of the file cards and six core-to-delta distances of the search card--prescribed value), or a value "0" as the sub-pattern difference, and

said card difference calculating means determines a sum obtained by adding all the pattern level feature differences of the respective fingers calculated by said fingerprint pattern level feature difference calculating means and the sub-pattern level feature differences of the respective fingers obtained by said fingerprint sub-pattern level feature difference calculating means as the card difference.

6. The tenprint card selector as set forth in claim 2, wherein

said fingerprint pattern level feature difference calculating means

receives the pattern level feature types of the fingerprint images of said file cards and said search card and receives the pattern level feature confidence rating to insure the correctness of the pattern level feature types of the fingerprint images of said file cards and said search card, and

determines a product of a pattern level feature confidence rating of the fingerprint images of said file cards and a pattern level feature confidence rating of the fingerprint images of said search card as a magnitude and a value with its code reversed as the pattern level feature difference when the pattern level feature types of the fingerprint images of said file cards and the pattern level feature types of the fingerprint images of said search card are identical, determines a product of a pattern level feature confidence rating of the fingerprint images of said file cards and a pattern level feature confidence rating of the fingerprint images of said search card as the pattern level feature difference when the pattern level feature types of the fingerprint images of said file cards and the pattern level feature types of the fingerprint images of said search card do not match, and determines the value "0" as the pattern level feature difference when at least either of the pattern level feature types is unclassifiable;

said fingerprint sub-pattern calculating means

receives said sub-pattern level information containing position coordinates of the singular points on the fingerprint images of said file cards and said search card, and also receives the pattern level feature types of the fingerprint images of said file cards and said search card and the sub-pattern level information confidence rating to insure the correctness of the sub-pattern level information concerning the fingerprint images of said file cards and said search card, and

calculates said distance between singular points to determine as the sub-pattern level feature and calculates a sub-pattern level feature confidence rating product by multiplying the sub-pattern level information confidence rating concerning the singular points used to calculate said sub-pattern level feature;

said fingerprint sub-pattern level feature difference calculating means,

when the pattern level feature types of the fingerprint images on said file cards and the pattern level feature types of the fingerprint images on the search card are loops and identical, whorls and identical, or others, determines (a value obtained by multiplying the absolute value of a difference between the core-to-delta distance of the file cards and the core-to-delta distance of the search card by a sub-pattern level feature confidence rating product--prescribed value), (average value obtained by multiplying the absolute values of differences between six core-to-delta distances of the file cards and six core-to-delta distances of the search card by the sub-pattern level feature confidence rating product--prescribed value), or a value "0" as the sub-pattern level feature difference; and

said card difference calculating means

calculates a first calculation value which is a sum of the pattern level feature differences of the respective fingers obtained by said fingerprint pattern level feature difference calculating means and a second calculation value which is a quotient obtained by dividing a sum of the sub-pattern level feature differences of the respective fingers obtained by said fingerprint sub-pattern level feature difference calculating means by a sum of the sub-pattern level feature confidence rating products of the respective fingers obtained by said fingerprint sub-pattern level feature difference calculating means, and determines a sum of said first calculation value and said second calculation value as the card difference.

12. The tenprint card selector as set forth in claim 2, wherein the card difference calculating unit calculates the card difference Dc according to Dc=(f(a_i D_pi)+g(b_i D_si), wherein f and g are functions, a_i and b_i are constants, D_pi corresponds to said pattern level feature difference, and D_si corresponds to said sub-pattern level feature difference.

7. A tenprint card preselector for preselecting tenprint cards which are to be compared by selecting the tenprint cards having features similar to those of a sample from a file containing a plurality of tenprint cards before matching fingerprints by using tenprint cards, comprising:

a first fingerprint pattern level judging means for detecting the candidates for pattern level feature types which are applicable to fingerprint images of respective fingers collected on file cards and producing a set of candidates;

a second fingerprint pattern level judging means for detecting the candidates for pattern level feature types which are applicable to fingerprint images of respective fingers collected on a search card and producing a set of candidates;

a tenprint card selecting means for receiving the set of candidates for the pattern level feature types for one card each from said first fingerprint pattern level feature judging means and said second fingerprint pattern level judging means and selecting the file cards which are to be compared with the search card;

said tenprint card selecting means comprising;

a fingerprint pattern level feature difference calculating means for receiving and comparing a set of candidates for pattern level feature types of fingerprint images on said file cards produced by said first fingerprint pattern level feature judging means and a set of candidates for pattern level feature types of fingerprint images on said search card produced by said second fingerprint pattern level feature judging means and calculating a pattern level feature difference to show the degree of similarity between the sets of candidates,

a card difference calculating means for calculating a card difference showing similarity of the features of the fingerprint images shown on said file cards and said search card based on said pattern level feature difference of the fingerprint images of all fingers shown on said file cards and said search card, and

a judging means for comparing the card difference determined by said card difference calculating means with a predetermined card difference threshold value and judging whether said file card shall be selected to be further checked with said search card,

wherein when the comparison of the sets of candidates for the pattern level feature types of said file cards and said search card results in that both sets of candidates are similar or at least either of the pattern level feature types is unidentified, that the sets of candidates do not have any common pattern level feature type, or that a plurality of candidates in the sets of candidates have the same pattern level feature types, said fingerprint pattern level feature difference calculating means determines the value of the pattern level feature difference to respectively be a minimum value, a maximum value, and a value appropriately determined between said minimum and maximum values, and

wherein said card difference calculating means determines a sum of all the pattern level feature differences of the respective fingers calculated by said fingerprint pattern level feature difference calculating means as the card difference.

8. A tenprint card preselector for preselecting tenprint cards which are to be compared by selecting the ten pin cards having features similar to those of a sample from a file containing a plurality of tenprint cards before matching fingerprints by using the tenprint cards, comprising:

a first fingerprint pattern level feature judging means for judging applicable pattern level feature types with respect to fingerprint images of respective fingers collected on file cards;

a first sub-pattern level feature detecting means for detecting sub-pattern level information of said fingerprint images based on the fingerprint images of respective fingers collected on said file cards,

a second fingerprint pattern level feature judging means for detecting candidates for applicable pattern level feature types with respect to fingerprint images of respective fingers collected on a search card and producing a set of candidates,

a second sub-pattern level feature detecting means for detecting sub-pattern level information of said fingerprint images based on the fingerprint images of respective fingers collected on said search card; and

a tenprint card selecting means for receiving the set of candidates for the pattern level feature types for one card each from said first fingerprint pattern level feature judging means and said second fingerprint pattern level feature judging means, receiving said sub-pattern level information of tenprint cards corresponding to the set of candidates received from said first sub-pattern level feature detecting means and said second sub-pattern level feature detecting means, and selecting the file cards which are to be compared with the search card;

said tenprint card selecting means comprising;

a fingerprint pattern level feature difference calculating means for receiving and comparing pattern level feature types of fingerprint images on said file cards judged by said first fingerprint pattern level feature judging means and pattern level feature types of fingerprint images on said search card judged by said second fingerprint pattern level feature judging means and calculating a pattern level feature difference to show whether or not said pattern level feature types are identical mutually,

a fingerprint sub-pattern level feature calculating means for receiving sub-pattern level information on the fingerprint images shown on said file cards and sub-pattern level information on the fingerprint images shown on said search card and calculating a sub-pattern level feature showing secondary features of said fingerprint images,

a fingerprint sub-pattern level feature difference calculating means for receiving and comparing the sub-pattern level feature of the file cards and the sub-pattern level feature of the search card determined by said fingerprint sub-pattern level feature calculating means and also receiving and referring to the pattern level feature types of the fingerprint images shown on said file cards and said search card, and calculating a sub-pattern level feature difference which shows the degree of similarity between said sub-pattern level features mutually,

a card difference calculating means for calculating a card difference showing similarity of the features of the fingerprint images shown on said file cards and said search card based on said pattern level feature difference of the fingerprint images of all fingers shown on said file cards and said search card and said sub-pattern level feature difference of the fingerprint images of all fingers shown on said file cards and said search card, and

a judging means for comparing the card difference determined by said card difference calculating means with a predetermined card difference threshold value and judging whether said file card shall be selected to be further checked with said search card.

9. The tenprint card preselector as set forth in claim 8, wherein

when the comparison of the pattern level feature types of said file cards and said search card results in that both pattern level feature types are identical or at least either of the pattern level feature types is unclassifiable, or that the pattern level feature types do not match, said fingerprint pattern level feature difference calculating means determines the value of the pattern level feature difference to be a minimum value, or a maximum value,

said fingerprint sub-pattern level feature calculating means receives said sub-pattern level information containing a position coordinate of the singular points on the fingerprint images of said file cards and said search card to calculate a distance between said singular points and determines it as the sub-pattern level feature,

when the pattern level feature types of the fingerprint images on said file cards and the pattern level feature types of the fingerprint images on the search card are identical, said fingerprint sub-pattern level feature difference calculating means calculates said sub-pattern level feature difference based on the positional difference of singular points which are sub-pattern level features according to a calculating formula determined according to said pattern level feature types, and

said card difference calculating means determines a sum obtained by adding all the pattern level feature differences of the respective fingers calculated by said fingerprint pattern level feature difference calculating means and the sub-pattern level feature differences of the respective fingers obtained by said fingerprint sub-pattern level feature difference calculating means as the card difference.

10. The tenprint card preselector as set forth in claim 8, wherein

when the comparison of the pattern level feature types of said file cards and said search card results in that both pattern level feature types are identical or at least either of the pattern level feature types is unclassifiable, or that the pattern level feature types do not match, said fingerprint pattern level feature difference calculating means determines the value of the pattern level feature difference to be a minimum value, or a maximum value,

said fingerprint sub-pattern level feature calculating means receives said sub-pattern level information containing a position coordinate of the singular points on the fingerprint images of said file cards and said search card to calculate a distance between said singular points and determines it as the sub-pattern level feature,

when the pattern level feature types of the fingerprint images on said file cards and the pattern level feature types of the fingerprint images on the search card are loops and identical, whorls and identical, or others, said fingerprint pattern level feature difference calculating means determines (the absolute value of a difference between the core-to-delta distance of the file cards and the core-to-delta distance of the search card--prescribed value), (average value of the absolute values of differences between six core-to-delta distances of the file cards and six core-to-delta distances of the search card--prescribed value), or a value "0" as the sub-pattern level feature difference, and

said card difference calculating means determines a sum obtained by adding all the pattern level feature differences of the respective fingers calculated by said fingerprint pattern level feature difference calculating means and the sub-pattern level feature differences of the respective fingers obtained by said fingerprint sub-pattern level feature difference calculating means as the card difference.

11. The tenprint card preselector as set forth in claim 8, wherein

said fingerprint pattern level feature difference calculating means

receives the pattern level feature types of the fingerprint images of said file cards and said search card and receives the pattern level feature confidence rating to insure the correctness of the pattern level feature types of the fingerprint images of said file cards and said search card, and

determines a product of -1, a pattern level feature confidence rating of the fingerprint images of said file cards and a pattern level feature confidence rating of the fingerprint images of said search card as the pattern level feature difference when the pattern level feature types of the fingerprint images of said file cards and the pattern level feature types of the fingerprint images of said search card are identical, determines a product of a pattern level feature confidence rating of the fingerprint images of said file cards and a pattern level feature confidence rating of the fingerprint images of said search card as the pattern level feature difference when the pattern level feature types of the fingerprint images of said file cards and the pattern level feature types of the fingerprint images of said search card do not match, and determines the value "0" as the pattern level feature difference when at least either of the pattern level feature types is unclassifiable;

said fingerprint sub-pattern calculating means

receives said sub-pattern level information containing position coordinates of the singular points on the fingerprint images of said file cards and said search card, and also receives the pattern level feature types of the fingerprint images of said file cards and said search card and the sub-pattern level information confidence rating to insure the correctness of the sub-pattern level information concerning the fingerprint images of said file cards and said search card, and

calculates said distance between singular points to determine as the sub-pattern level feature and calculates a sub-pattern level feature confidence rating product by multiplying the sub-pattern level information confidence rating concerning the singular points used to calculate said sub-pattern level feature;

said fingerprint sub-pattern level feature difference calculating means,

when the pattern level feature types of the fingerprint images on said file cards and the pattern level feature types of the fingerprint images on the search card are loops and identical, whorls and identical, or others, determines (a value obtained by multiplying the absolute value of a difference between the core-to-delta distance of the file cards and the core-to-delta distance of the search card by a sub-pattern level feature confidence rating product--prescribed value), (average value obtained by multiplying the absolute values of differences between six core-to-delta distances of the file cards and six core-to-delta distances of the search card by the sub-pattern level feature confidence rating product--prescribed value), or a value "0" as the sub-pattern level feature difference; and

said card difference calculating means

calculates a first calculation value which is a sum of the pattern level feature differences of the respective fingers obtained by said fingerprint pattern level feature difference calculating means and a second calculation value which is a quotient obtained by dividing a sum of the sub-pattern level feature differences of the respective fingers obtained by said fingerprint sub-pattern level feature difference calculating means by a sum of the sub-pattern level feature confidence rating products of the respective fingers obtained by said fingerprint sub-pattern level feature difference calculating means, and determines a sum of said first calculation value and said second calculation value as the card difference.