Analysis of Fingerprint Minutiae to Form Fingerprint Identifier

Detailed human fingerprints, almost unique, are difficult to change and are permanent on an individual's life, making them suitable as long-term signs of human identity. They may be employed by the police or other authorities to identify individuals who wish to conceal their identity, or identify incapacitated or deceased persons and therefore cannot identify them, as in the aftermath of a natural disaster. Fingerprints images are very important data type due to wide applications requiring this type, so extraction a fingerprint identifier is a vital issue. In this paper we will analyse the fingerprints images in order to extract minutiae from the images, these minutiae will be used to construct the fingerprint identifier, the proposed procedure will be implemented and tested to ensure that the procedure generates a simple and unique identifier, which can be easily used to recognize the fingerprint in any recognition system. Keywords— Fingerprint, digital image, minutiae, ridge ending, bifurcation, identifier, Euclidean distance.


I. INTRODUCTION
The human fingerprint is the impression left by the friction of the human finger. Recovering partial fingerprints from the crime scene is an important method of forensic science. The moisture and grease in the finger causes fingerprints on surfaces such as glass or metal. Deliberate impressions of entire fingerprints can be obtained by ink or other materials transferred from the tops of friction edges on the skin to a smooth surface such as paper. Fingerprint records typically contain impres-sions from the panel on the last joint of the fingers and thumbs, although fingerprint cards usually record parts of the areas of the lower finger joint [1], [2].
Detailed human fingerprints, almost unique, are difficult to change and are permanent on an individual's life, making them suitable as long-term signs of human identity. They may be employed by the police or other authorities to identify individuals who wish to conceal their identity, or identify incapacitated or de-ceased persons and therefore cannot identify them, as in the aftermath of a natural disaster [3], [4].
To simplify the process of fingerprint identification and make efficient and accurate, we have to represent the fingerprint by a small in size array of features [21[, [22], which can be easily used in the process of identification [14], [15], [16]. Many methods were used to extract image features to be used as an identifier [17], [18], the extracted features must be characterized with the following: The features for each image must be unique to maximize the identification system accuracy. -Must have a small size to reduce the needed memory space.
In this paper we will analyze and discuss the minutiae method of features extraction, concentrating our attention on ridge ending and bifurcations.

II. FINGERPRINT CHARACTERISTICS
Human fingerprint has a unique structure, and each fingerprint is composed of various objects, the repetition of each object, and the continents of each object are unique and differ from one person to another, so we can based on these objects to create a unique identifier for each fingerprint [1], [2].
Each fingerprint consists of a set of minutiae, minutiae has different shapes as shown in figure 2, these shapes can be considered as an objects which we have to detect in order to build a fingerprint identifier.  To detect and extract any minutiae we have to follow the following steps: -Define a 3 by 3 mask, which we will call a classifier number (CN), this mask will be anded with each pixel and its neighbors, the sum of ones in the result will point to CN as shown in figure 4.

III. PROCEDURAL ANALYSIS OF MINUTIAE EXTRACTION
The procedural analysis required to extract various fingerprint minutiae can be performed as shown in figure 6 applying the following steps: To minimize the number of points in each minutiae and the extraction time, we can take a take a selected segment of the image with smaller size (for example 100x100 pixels). Figure 7 shows an example of thinned image and the extracted minutiae.   Taking the thinned original image as an input to detect and extract minutiae will lead to extracting minutiae with large number of points, thus the size of the obtained minutiae will be large and the extraction time will be also big, figure 11 shows some samples of treated fingerprint, while table 1 shows the extracted minutiae:  In this paper research we will focus only on ridge ending and bifurcation types of minutiae, they are sufficient to create a unique identifier for a fingerprint, as will be shown later in this part. Table 2 shows the first 10 points coordinates of the ridge ending minutiae, while table 3 shows the first 12 points coordinates of bifurcation of the first sample of finger print:   From table 4 we can see that average extraction time is significantly high, also the number of points in each minutiae is also big, which leads to a huge number of extracted points.
The minimize the negative effects of the above mentioned disadvantages, we can use a selected segment of the fingerprint with a small fixed size, table 5 shows the extracted minutiae for the above used fingerprints with a 100x100 pixel segment  From table 5 we can see the following facts: -The extraction time was significantly reduced.
-The number of points in each minutia also was reduced, reducing the memory space required to store the points. -The number of points for each finger print is unique. -The points coordinate values for each minutia are also unique. To minimize the size of the fingerprint identifier size we can form the identifier from the following components: -Number of points in the ridge ending minutiae.
-Euclidean distance of the ridge ending point [20] which can be calculated applying the following formula: Where p and q are the points coordinates. -Number of points in the bifurcation.
-Euclidean distance of bifurcation points. Table 6 shows the extracted and the calculated components of fingerprint identifiers:

V. CONCLUSIONS
A simple, accurate and efficient procedure for fingerprint identifier extraction was proposed, tested and implemented. The obtained exper-imental showed the following facts: -The procedure requires small amount of time to extract minutiae. -The memory space required is small. -It is better to use a fingerprint seg-ment to minimize the extraction time and the memory space. -We can use Euclidean distance to re-place the points coordinates in the identifier, making the identifier small-er and easier to handle. -The obtained identifier for each fin-gerprint is unique, and it can be accu-rately used to retrieve the fingerprint in a recognition system