A Soft Computing Technique for Improving the Fidelity of Thumbprints Based Identification Systems

Published Online July 2016 in MECS

Soft Computation techniques use few uncertain aspects to produce near to certain results. Talking about human and natural traits which can be used in an identification system, uncertain disparities and partial truth comes out along with results. Continuous perusal on thumbprints images have given few more untouched regions that can help to form better and efficient thumbprint based identification systems. A thumbprint is the biometric characteristic which is defined as oval impression and it consists of arch, left-right loops, trench and whorl etc.. Different fields in soft computing have their own importance like neural network which widely exists in many identification systems. The genetic algorithm is encoded with the help of decision variables. The fuzzy logic is also an effective tool that deals with the vagueness and uncertain aspects of the system with its major Classification: Fuzzy Set Theory and Fuzzy System. The biometric system involves several human traits like face shape; square recessive; oval dominant; Hair: Curly, Wavy, Straight; Eye shape: round recessive, Almond dominant; and Earlobes shape: attached recessive, Free lobe [10]. But, unfortunately we cannot use them for identification purpose because if we form clusters or sets based on these traits then a huge set of people will fall under single category. Therefore, human traits which we use in system for verification should follow necessary criterion like Amenity (every subject should have common attribute and do not change group to group or person to person), disjunction (information shared by attributes should not be same or must vary from person to person) and indelibility (do not deplete or change over time and it should be quantifiable)

In few cases like dead bodies where finger prints are not an option and even dental evidence is difficult. In this situation palatal patterns can be taken into consideration. Palatal patterns show patterns like finger prints which bring it into category of traits that can be used for human identification [33]. The iris image also shows textures and patterns that can be quantified and used in effective identification of human but with few bottlenecks: surgery can change its original pattern [3]. The different traits which are used for personal identification are presented in fig. 1.

In general sense we can say that the proposed system should be robust, secure and can be relied over long period of time with negligible false acceptance rates. The fraternal twins are easily verifiable but identical twins identification is a challenging task. With increasing twin ratio every year we need to upgrade existing systems [11].

Fig.1. Identification system using different traits for personal identification.

In this paper we proposed a new refined technique that will help our existing thumbprint based identification systems to identify individual more accurately and it will also work for identical twins identification. The proposed technique of this paper is the extension of existing thumbprint based biometric identification systems which are effective in terms of space and efficient in terms of time with improved fidelity. In this paper soft computing technique along with 2D transformation is proposed to discover a better and flexible identification system.

In the next section (section 2) we have discussed literature review of existing systems. The section 3 describes mathematical model of the proposed system. In section 4 we have proposed new algorithms and model which uses soft computing techniques for improving the fidelity of thumbprint based systems. The section 5 analyses the performance of proposed system and the results of proposed algorithm is compared with other techniques in this section. The section 6 concludes the outcomes of this paper.

• II.    Literature Review