Abstract—An efficient iris segmentation method based on analyzing the local entropy characteristic of the iris image, is proposed in this paper and the strength and weaknesses of the method are analyzed for practical purposes. The method shows special strength in providing designers with an adequate degree of freedom in choosing the proper sections of the iris for their application purposes. Keywords—Iris segmentation, entropy, biocryptosystem, biometric identification. I. INTRODUCTION Y the advance of the concept of the electronic world village into human societies from one side and the increasing need for restricting access to confidential data from the other, the need for introduction of modern and effective authentication schemes are becoming more necessary. In this case biometrics shows a distinct and unique presence. The fact that biometric data are the distinct character of the individual and are seldom or never changed because of external effects, puts biometric in an advantageous position compared with typical nonpersonal schemes based on PIN codes and passwords. Among the present biometric topics human iris has a distinctively promising character. Its special location in the eye structure makes, unlike e.g fingerprints, capturing unwanted copies harder and more complicated, also iris shows distinctive characters related to its shape, form and specially texture and surface features, which in return delivers a high amount of relatively reliable data for according application processing. Conventional iris based authentication schemes use a double step process for the identification of individuals. In the first step a reference iris image is captured from the owner and after performing image processing algorithms the obtained data is stored in a database, In the 2 nd step and during the authentication process a sample image is captured from the claimer, which goes through the same processing algorithms. Afterwards by making comparison between the previously stored data and the present one the validity or invalidity of the claim is deduced and accordingly the former leads to authentication and the later to rejection. However to store the database of the iris reference images in someplace always results in the threat of the database being hacked and therefore the entire security vanished. The first solution that seems reasonable is to somehow omit the necessity for storing image database and finding another alternative for fulfilling the authentication process. Hao, Anderson and Daugman [3] have proposed an effective method based on combining cryptography and biometrics for omitting the need for storing the reference data in the system structure, However as it is passively shown in [3] the structure suffers much from the presence of the burst noise which in return asks for sacrificing a part of system capacity for suppressing the unwanted effects of the burst noise. Also from the view of the authors of this paper it may not be a proper idea to choose the entire surface of the iris image for the key strengthening purposes as the presence of high level of correlation in iris data can result in dependent and therefore weak key strings. The method which is presented in this paper and which shall be called “local entropy method” in the rest of the paper, is primarily designed as an alternative for classical iris segmentation methods like the Hough transform. However, as a secondary application it shall also be shown that the method provides a means for choosing the best regions within the iris, for biocrypto purposes. Moreover, the results of the experiments show that combined with proper image processing methods, the method shows a strong resistance against the effects of burst errors. The remainder of the paper is organized as follows. After the introduction in section 2 the mathematical background of local entropy method is explained. In section 3 the necessary image processing methods required for preparing the image for local entropy method are described. In section 4 the potential applications of the method are discussed and compared with similar cases. Finally the conclusions are given in section 5. II. DEFINITION OF LOCAL ENTROPY METHOD A. What is Local Entropy? According to Shannon’s 2 nd theorem [1] if the event i occurs from a set of valid events, with the probability i p the amount of uncertainty related to the event is equal to: ) / )( ( log 2 Symbol bits p H i i − = (1) And also the amount of the uncertainty that the source of the events generates is equal to: ) ( )) ( log ( 2 bits p p H i i ∑ − = (2) From equation (2) it can be seen that the highest amount of uncertainty from an information source is realized when the output symbols of the source are equally probable. The idea behind local entropy method is to divide the processed image into separate regions and then to analyze An Efficient Segmentation Method Based on Local Entropy Characteristics of Iris Biometrics Ali Shojaee Bakhtiari, Ali Asghar Beheshti Shirazi, and Amir Sepasi Zahmati B INTERNATIONAL JOURNAL OF BIOMEDICAL SCIENCES VOLUME 2 NUMBER 3 2007 ISSN 1306-1216 IJBS VOLUME 2 NUMBER 3 2007 ISSN 1306-1216 195 © 2007 WASET.ORG