Fingerprint processing system capable of detecting a core of a fingerprint image by statistically processing parameters

• US 5,040,224 A
• Filed: 04/24/1989
• Issued: 08/13/1991
• Est. Priority Date: 04/23/1988
• Status: Expired due to Term

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Hara, Masanori
• Primary Examiner(s): Moore, David K.
• Assistant Examiner(s): Couso, Jose L.
• Application Number:
  US07/342,047
• Time in Patent Office: 841 Days
• Field of Search: 382/4, 382/5, 382/39
• US Class Current: 382/124
• CPC Class Codes: G06V 40/1347 Preprocessing; Feature extraction

Abstract

In a fingerprint processing system for use in detecting a core position of a fingerprint image quantized into a plurality of picture elements, statistical processing is carried out in relation to parameters which are variable in dependency upon the core position and directions and curvatures of the picture elements preliminarily detected to specify ridges of the fingerprint image. The parameters may be existence probabilities of the core position that are determined for the directions and the curvatures. Such existence probabilities are successively calculated at each picture element and adjacent picture elements to be summed up and to detect a maximum one of the existence probabilities. A position which has the maximum existence probability is judged as the core position. Alternatively, the parameters may be differences between practical curvatures extracted from the fingerprint image and reference curvatures defining a reference curve. A minimum one of the differences is statistically detected with reference to various kinds of the reference curves. In addition, the directions and the curvatures of the picture elements may be determined by deriving the density levels of the picture elements along a plurality of arcuate lines specified by line directions and line curvatures.

Claims

1. A fingerprint processing system for use in detecting a position of a core in a fingerprint image which has a plurality of ridges, each having a curvature and a direction, said ridges being quantized into picture elements together with their background, said system comprising a direction and curvature memory having a plurality of memory addresses in correspondence to said picture elements, respectively, for memorizing curvature signals and direction signals and a processing unit coupled to said curvature and direction memory for successively processing said curvature signals and said direction signals to detect said position, said curvature and said direction signals being extracted from said picture elements to represent the curvatures and the directions, respectively, the improvement wherein:

said processing unit comprises;

statistically processing means coupled to said curvature and direction memory for successively and statistically processing said curvature and said direction signals at said picture elements to calculate parameters defined by existence probabilities of said position which are calculated in correspondence to combinations of said directions and said curvatures of the ridges; and

position determining means connected to said statistically processing means for determining said position from the parameters;

said statistically processing means comprising;

a probability thesaurus having a plurality of thesaurus addresses corresponding to combinations of said curvatures and said directions of the ridges for memorizing, in each of said thesaurus addresses, a relative existence probability set of said position which is calculated in relation to a normalized reference picture element and a plurality of adjacent normalized picture elements adjacent to said normalized reference picture element to produce said relative existence probability set, said relative existence probability set being produced one at a time in response to said combinations;

accessing means coupled to said curvature and direction memory and said probability thesaurus and successively supplied with the curvature and the direction signals memorized at each of said memory addresses for accessing said probability thesaurus to extract the relative existence probability set from the thesaurus address corresponding to said combination of the curvature and the direction signals and to produce the relative existence probability set together with a set of location signals representing, on said fingerprint image, locations of the picture elements given said relative existence probabilities, respectively;

existence probability memory means for memorizing practical existence probabilities of said position at the picture elements on said fingerprint image to produce practical existence probability signals representative of the memorized practical existence probabilities, respectively; and

calculation circuit means coupled to said accessing means and said existence probability memory means for successively and statistically calculating, as said parameters, the practical existence probabilities for said picture elements on said fingerprint image, said practical existence probabilities having a maximum value at the location indicated by the selected one of said location signals;

said position determining means being coupled to said existence probability memory means to determine said position of the core from said parameters memorized in said existence probability memory means.

2. A fingerprint processing system for use in detecting a position of a core in a fingerprint image which has a plurality of ridges, each having a curvature and a direction, said ridges being quantized into picture elements together with their background, said system comprising a direction and curvature memory having a plurality of memory addresses in correspondence to said picture elements, respectively, for memorizing curvature signals and direction signals and a processing unit coupled to said curvature and direction memory for successively processing said curvature signals and said direction signals to detect said position, said curvature and said direction signals being extracted from said picture elements to represent the curvatures and the directions, respectively, the improvement wherein:

said processing unit comprises;

statistically processing means coupled to said curvature and direction memory for successively and statistically processing said curvature and said direction signals at said picture elements to calculate parameters variable in dependency upon said position and upon said curvatures and said directions of the ridges; and

position determining means connected to said statistically processing means for determining said position from the parameters;

said fingerprint image being impressed on a plane defined by a coordinate system having a first axis and a second axis orthogonal to said first axis, said ridges having convex curves relative to said first axis, wherein said statistically processing means comprises;

curvature detecting means coupled to said direction and curvature memory for detecting a maximum one of the curvatures of the picture elements arranged along said first axis at every one of coordinate positions placed along said second axis to successively produce the maximum ones of the curvatures and the corresponding positions as detected curvatures and detected positions;

a curvature thesaurus for memorizing a plurality of reference curves representative of specific fingerprint images, each of said reference curves being specified by a plurality of reference curvatures and the corresponding curvature positions placed along said second axis, together with a provisional coordinate position of the core related to each of said reference curvature curves; and

calculating means coupled to said curvature detecting means and said curvature thesaurus for statistically calculating errors between the detected curvatures and said reference curvatures with reference to said detected positions and the curvature positions by changing said provisional coordinate position to another, said reference curvatures and said provisional coordinate position specifying a selected one of said reference curves, said calculating means being for producing, as said parameters, a minimum one of said errors and the corresponding provisional position that correspond to said minimum one of the errors, said minimum one of the errors and the corresponding provisional coordinate position being calculated for every one of said reference curves as calculated minimum errors and calculated provisional coordinate positions;

said position determining means selecting a minimal one of said calculated minimum errors to determine as said position one of the calculated provisional coordinate positions that corresponds to said minimal one of said calculated minimum errors.