minietua algorithm

Upload: aashi-jain

Post on 05-Apr-2018

218 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/31/2019 Minietua Algorithm

    1/16

    Minietua Algorithm

  • 7/31/2019 Minietua Algorithm

    2/16

    What is Minietua Matching Algorithm?

    It is algorithm to extracting minietua from theFingerprints and match them to show if bothfingerprint belong to same person or not.

  • 7/31/2019 Minietua Algorithm

    3/16

    Fingerprint Description

    The lines that flow in various pattern acrossFingerprint are called ridges and spaces betweenridges are called valley.

    These ridges are compared between one fingerprintand other when matching.

    Comparison of all ridges is a lengthy task. Henceinstead of ridges minietua are matched.

  • 7/31/2019 Minietua Algorithm

    4/16

    Ridge

  • 7/31/2019 Minietua Algorithm

    5/16

    What actually is minietua?

    Minutiae are the special points of interest in aFingerprint such as ridge ending or ridge bifurcation.

    Ridge Ending RidgeBifurcation/Intersection

  • 7/31/2019 Minietua Algorithm

    6/16

    Types of minietua

    Ridge Ending.

    Ridge Intersection.

    Ridge Bifurcation.

    Lake

  • 7/31/2019 Minietua Algorithm

    7/16

    Steps in Minietua Matching

    Step 1 : Binarization.

    Step 2 : Smoothening.

    Step 3 : Thinning (Skeletonization).

    Step 4 : Extract Minutiae.Step 5 : Match Minutiae from Database.

  • 7/31/2019 Minietua Algorithm

    8/16

    Binarization

    Binarization is the process to convert a gray scale imageinto a Binary Image to improve the contrast betweenridges and valleys in the fingerprint image.

    How to Binarize image.

    Calculate Gray level of each pixel by taking mean of RGB values atthose pixel.

    Calculate the Global mean by taking mean of all above calculatedvalues.

    Calculate the local mean of small regions of the image.

    If the local mean is greater than the global mean then local greymean is calculated by formula localmean = 0.75*global mean + 0.25*localmean

    Now each pixel is compared with local mean if its value is greaterthan local mean then it is set to binary one else it is set to binaryzero.

  • 7/31/2019 Minietua Algorithm

    9/16

    Original FingerprintImage

    Binary FingerprintImage

  • 7/31/2019 Minietua Algorithm

    10/16

    Smoothen Image

    Image is smoothened by processing the image withthe filter.

    Here, Mean Filter is used with properties

    filter ={0.0625 0.1250 0.06250.1250 0.2500 0.12500.0625 0.1250 0.0625 }

    Every pixel is calculated for weight of its neighbour

    by adding corresponding value of binary pixel.

    If weight is greater than 0.5 than change the pixelvalue to binary one.

  • 7/31/2019 Minietua Algorithm

    11/16

    BinaryFingerprint

    SmoothenFingerprint

  • 7/31/2019 Minietua Algorithm

    12/16

    Skeletonization (Thinning)

    Sekletonization is done through Zhang-Suen algorithm on binary image from laststep.

    This algorithm is made by two subiterations. In the first one, a pixel I(i,j) isdeleted if the following condition aresatisfied: It has at least two black neighbors and not more

    than six.

    At least one of I(i,j+1), I(i-1,j), and I(i,j-1) are white.

    At least one of I(i-1,j), I(i+1,j), and I(i,j-1) are white.

  • 7/31/2019 Minietua Algorithm

    13/16

    In the second Iteration. It has at least two black neighbors and not more than

    six.

    At least one of I(i-1,j), I(i,j+1), and I(i+1,j) are white. At least one of I(i,j+1), I(i+1,j), and I(i,j-1) are white.

  • 7/31/2019 Minietua Algorithm

    14/16

    Minietua Extraction

    To extract minietua we check for neighborof each pixel

    For end points

  • 7/31/2019 Minietua Algorithm

    15/16

    For Bifurcation points

  • 7/31/2019 Minietua Algorithm

    16/16

    Minietua Matching

    The various minietua from last step to theminietua stored in the database arematched and checked for the matched

    percentage if more than 70% of theseminietua are matched than fingerprint ismatched.