Signaler

Partager

•0 j'aime•74 vues

As a part of a growing information society, security and the authentication of individuals become nowadays more than ever an asset of great significance in almost every field. Iris recognition system provides identification and verification automatically of an individual based on characteristics and unique features in iris structure. Accurate iris recognition system based on iris segmentation method and how localized the inner and outer iris boundaries that can be damaged by irrelevant parts such as eyelashes and eyelid, to achieve this aim, the proposed method applied using [Iris Segmentation Based on Brightness Correction ISBC] to edge detection then circle distribution "CD" transformation on an eye image passed through preprocessing operations. The proposed iris normalization is done by using the important information that resulted from iris segmentation such as center and radius of iris to convert iris region in the original image from the Cartesian coordinates (x, y) to the normalized polar coordinates (r, θ). The proposed approach tested conducted on the iris CASIA (Chinese Academy of Science and Institute of Automation) data set (CASIA v1.0 and CASIA v4.0-interval ) iris image database and the results indicated that proposed approach has 100% accuracy rate with (CASIA v1.0), and has 100% accuracy rate with (CASIA v4.0- interval) .

•0 j'aime•74 vues

Signaler

Partager

As a part of a growing information society, security and the authentication of individuals become nowadays more than ever an asset of great significance in almost every field. Iris recognition system provides identification and verification automatically of an individual based on characteristics and unique features in iris structure. Accurate iris recognition system based on iris segmentation method and how localized the inner and outer iris boundaries that can be damaged by irrelevant parts such as eyelashes and eyelid, to achieve this aim, the proposed method applied using [Iris Segmentation Based on Brightness Correction ISBC] to edge detection then circle distribution "CD" transformation on an eye image passed through preprocessing operations. The proposed iris normalization is done by using the important information that resulted from iris segmentation such as center and radius of iris to convert iris region in the original image from the Cartesian coordinates (x, y) to the normalized polar coordinates (r, θ). The proposed approach tested conducted on the iris CASIA (Chinese Academy of Science and Institute of Automation) data set (CASIA v1.0 and CASIA v4.0-interval ) iris image database and the results indicated that proposed approach has 100% accuracy rate with (CASIA v1.0), and has 100% accuracy rate with (CASIA v4.0- interval) .

- 1. Enhanced Approach to Iris Normalization Using Circular Distribution for Iris Recognition system Taha Mohammad Hassan Department of Computer Science, College of Science, University of Diyala Diyala, Iraq dr.tahamh@sciences.uodiyala.edu.iq Naji Mutar Sahib Department of Computer Sciences, College of Science, University of Diyala, Diyala, Iraq naji_e2006@yahoo.com Abdul Salam Hassan Abbas Department of Computer Science, College of Science, University of Diyala Diyala, Iraq slamalqrtghwly@gmail.com Abstract— As a part of a growing information society, security and the authentication of individuals become nowadays more than ever an asset of great significance in almost every field. Iris recognition system provides identification and verification automatically of an individual based on characteristics and unique features in iris structure. Accurate iris recognition system based on iris segmentation method and how localized the inner and outer iris boundaries that can be damaged by irrelevant parts such as eyelashes and eyelid, to achieve this aim, the proposed method applied using [Iris Segmentation Based on Brightness Correction ISBC] to edge detection then circle distribution "CD" transformation on an eye image passed through preprocessing operations. The proposed iris normalization is done by using the important information that resulted from iris segmentation such as center and radius of iris to convert iris region in the original image from the Cartesian coordinates (x, y) to the normalized polar coordinates (r, θ). The proposed approach tested conducted on the iris CASIA (Chinese Academy of Science and Institute of Automation) data set (CASIA v1.0 and CASIA v4.0-interval ) iris image database and the results indicated that proposed approach has 100% accuracy rate with (CASIA v1.0), and has 100% accuracy rate with (CASIA v4.0- interval) . Keywords-Iris Normalization-CASIA(V1.0,V4.0-interval) database-Circular Distribution (CD). I. INTRODUCTION The word ―biometrics is derived from the Greek words bio (life) and metric (to measure) [1]. It is general term used alternatively to describe a characteristic or a process. As a characteristic it is measurable biological and behavioral characteristic that can be used for automated recognition. As a process it encompasses automated methods of recognizing an individual based on measurable biological and behavioral characteristic. Biometric identification may be preferred over traditional methods (e.g. passwords, smart-cards) because its information is virtually impossible to steal. Iris recognition has a very good balance of all the properties. A number of biometric characteristics are being used in various applications as Universality, Uniqueness, Permanence, Measurability, Performance, Acceptability, and Circumvention [2]. II. RELATED WORKS Once the segmentation module has estimated the iris’s boundary, the normalization module uses image registration technique to transform the iris texture from cartesian to polar coordinates. The process often called iris unwrapping, yields a rectangular entity that is used for subsequent processing [3], Iris Normalization has many algorithms such as : The John Daugman, 2004 [4], has applied a Rubber Sheet Model to map the sampled iris pixels from the Cartesian coordinates to the normalized Polar coordinates. It transforms the iris region from Cartesian coordinates (x, y) to Polar coordinates (r, θ) where r lies in the interval of [0, 1] and θ is in the range of [0, 2π]. However, this algorithm does not compensate for the rotation variance. I(x(r, θ ),y(r, θ))= I(r, θ ), where I(x, y) is the iris image, (x, y) are the Cartesian coordinates and (r, θ ) are the polar coordinates.The Wildes, 1997 [5], has proposed an image registration technique for normalizing iris textures. In this method, a newly acquired image, Ia(u, v) would be aligned with an image in the database, Id(u, v), that the comparison is performed. The alignment process is a transformation using a choice of mapping function, (U(x, y), V (x, y)) such that for all (x, y), the image intensity value at (x, y), in Ia is close to that at (x, y) in Id. 2 , , ,d ax y I x y I x u y v dxdy . (1) The Li Ma, 2002 [6], 2000[7], approach for Iris Normalization Li Ma [6] used Daugman’s Rubber Sheet Model and then Iris Image Enhancement and De-noising. The homogeneous rubber sheet model algorithm remaps each pixel in the localized iris region from the Cartesian coordinates to polar coordinates. Then Li Ma [6] maps the iris ring to a rectangular block of the texture of a fixed size anti-clockwise. The International Journal of Computer Science and Information Security (IJCSIS), Vol. 15, No. 9, September 2017 102 https://sites.google.com/site/ijcsis/ ISSN 1947-5500
- 2. normalized iris image still has low contrast and may have non- uniform illumination (lighting) caused by the position of light sources. Local histogram [7] analysis is applied to the normalized iris image to reduce the effect of not uniform illumination and obtain well-distributed texture image and noise is removed by filtering the image with a low-pass Gaussian filter. The Tisse [8], approach, the Rubber Sheet Model is used which remaps each point within the iris region to a pair of polar coordinates (p, θ) is used so the pupil changes in size X(p , θ) = (1-p) * XP(θ) + p *Xi(θ) (2) Y(p , θ) = (1-p )* Yp(θ) + p *Yi(θ) (3) Where Xp(θ), Yp(θ) and Xi(θ), Yi(θ) are the coordinates of the pupillary and limbic boundaries in the direction θ. Boles [9] proposed the normalization of images at the time of matching. When two images are considered, one image is considered as a reference image. The ratio of the maximum diameter of the iris in this image to that of the other image is calculated. This ratio is used to make the virtual circles on which data for feature extraction is picked up. The dimensions of the irises in the images are scaled to have the same constant diameter regardless of the original size in the image. III. OVERVIEW OF THE POPOSED APPROACH The proposed approach for iris normalization consists of three main stages: iris image acquisition, iris boundaries localization, and iris normalization that is shown in Figure 1. A. Iris Image Acquisition Stage This step is one of the most significant and deciding factors for obtaining a good result. A fine and clear image eliminates the process of noise removal and also helps in avoiding errors in calculation. In this case, computational errors are avoided due to the absence of reflections, and because the images have been taken from close proximity. This project uses the image provided CASIA-Version 1.0 and CASIA-interval Version 4.0. B. Preprocessing on Eye Imageris Image Acquisition Stage One of the major issues in the iris segmentation and recognition system is the preprocessing, which can result in accurate iris and pupil localization and thereby leads to excellent iris recognition system performance. In this work, the preprocessing on image consist of the four stages: convert the color image (if the image is color) to the grayscale image, brightness correction, convert to binary image, and enhanced by the median filter to remove some noise around pupil as shown in Figure. 2. Image acquisition Preprocessing Gray-Scale image Brightness correction Binary image Smoothing image Pupil Detection Many objects detection in binary image Choice a largest object as pupil Iris Outer Boundary Detection Split operation Enhancement operation Enhance with X-direction Enhance with Y-direction Enhance with Median filter Iris detection Iris isolation Iris Normalization Translate with Circular Distrebution Translate with Circular Distrebution Translate with Circular distribution Resize Iris Region Figure. 1: Block diagram of the proposed iris normalization using circular distribution. International Journal of Computer Science and Information Security (IJCSIS), Vol. 15, No. 9, September 2017 103 https://sites.google.com/site/ijcsis/ ISSN 1947-5500
- 3. Figure. 2. Preprocessing stages: (a) The input image, (b) Apply the brightness correction,(c)Apply the thresholding method (binary image ), (d) Removing noise in the image using (median filter) operation. In brightness correction two brightness values are choosing automatically depending on the average (mean) of the input (original) image: A - The first brightness (FB) value is an addition to the “grayscale image” - in most cases worth 20, except for some images that have eyelashes and eyelids very close to the pupil and are more than 20. B - The second brightness (SB) value is an addition to the “original image” most cases are worth 20, except for some images that are high brightness and are less than 20 or do not need to add brightness. Where the best two brightness (FB and SB) choosing to depend on: 1) Average of a grayscale image from 150 to 159 the best two brightness is [FB=20, SB=20]. 2) Average of a grayscale image from 160 to 169 the best two brightness is [FB=15, SB=15]. 3) Average of a grayscale image from 170 to 179 the best two brightness is [FB=10, SB=10]. 4) Average of a grayscale image from 180 to 189 the best two brightness is [FB=5, SB=5]. Average of a grayscale image from 190 and more above in these cases do not need to add any brightness correction. C. Pupil Detection In this paper, the inner boundary (pupil-iris boundary) is detected before the outer boundary (iris-sclera boundary), due to the fact that the pupil region is the darkest region in the eye image and can be detected easily. In addition, this can contribute to improving the accuracy and the speed of detecting the outer boundary as will be explained later. The pupil localization is a process to localize the pupil by transforming the gray scale eye image into a binary image using the threshold method depended on the stable threshold is 126. In this method, all pixels that have values Greater than the threshold are marked as edge points. In this method, there could exist some noise present in the binary image, due to other dark regions, such as eyelashes and eyelids. A filtering operation (median filter) applied to eliminate such noise. The pupil region is almost completely detected in the binary image after enhanced (median filter) image, then detect many objects contain true value as shown in Fig 3(b). After this operation, using a set of layers, each object detection that is putting in one of the layers as shown in Fig 3(c), and then apply correction on the values of each layer as shown in Fig 3(d).After many object detection and correction in the image, drawing each object finding, as shown in Fig 3(e). Then choose the large box as pupil object. Since it can be modeled as a box shape, the pupil is detected correctly by employing a modified to obtain the center coordinates and radius of the pupil box, as shown in Fig 3(f). a b c d T T T T T F T T T T T T T T F F T T T T T T F F T F F F T F F F T F F F T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T F F T F F F T T T F T T T T T T T T T T F T T T F F F T F F T T T T T T T T T T T T T T T T T T T T T T T T T T T T T T (a) (b) (c)(d) (e) (f) International Journal of Computer Science and Information Security (IJCSIS), Vol. 15, No. 9, September 2017 104 https://sites.google.com/site/ijcsis/ ISSN 1947-5500
- 4. Figure. 3. Pupil localization: (a) Enhanced binary image, (b) Many object detection, (c) Many layers using (each object put in the layer ), (d) Correct the value of objects in the layer, (e) drawing the objects, and (f)Choice large object as the pupil. Fig. 4-5, shows the correct detection of the inner boundary of the iris-pupil region, On CASIA version 1.0 and version 4.0-interval respectively. Fig. 4. Pupil localization stages in the CASIA version 1.0: (a) The input image, (b) Applying the increasing brightness of the image (c) Applying the thresholding method, (d) The output of the filtering operation (e) Detect many object, (f)Choice large object as the pupil. Fig. 5. Pupil localization stages in the CASIA version 4.0 interval: (a) The input image, (b) Applying the increasing brightness of the image (c) Applying the thresholding method, (d) The output of the filtering operation (e) Detect many object, (f)Choice large object in the pupil. D. Iris Outer Boundary Detection This stage is to find the outer iris boundary. it aims to find iris region, which is characterized as a color region that is surrounded by white color area (Sclera) from outside and black color area (pupil) from inside. This stage should be accomplished with high accuracy because it significantly affects the success of the system. To make the process simple and accurate. The proposed algorithm of iris outer boundary localization that includes six steps: split operation of an image, enhancing by X-direction, enhancing by Y-direction, applying a median filter, iris localization and fitting circle, and iris isolation as described in the Fig 6-7, on CASIA version 1.0 and version 4.0-interval respectively. The split operation: applied to find iris outer boundary separately, this operations calculates the average value (Avg) of the points extended along the inclined line in the “Gray Scale Image enhanced by brightness correction”, and then compare each point (P) along the inclined line with average value (Avg): if the value of (P) is below Avg then set (P) value to zero, else set (P) value to one, Fig 6-7(c). The Enhancing image with X – direction: this operation to remove the small gaps or pores (i.e., short runs of zeros or ones) which may found in the binary sequence of (P) in Y-direction; this step will lead to long run of zeros followed by long run of ones, Fig 8-9(d). The Enhancing image with Y – direction: this operation to remove the small gaps or pores (i.e., short runs of zeros or ones) which may found in the binary sequence of (P) in Y-direction; this step will lead to long run of zeros followed by long run of ones, Fig 6-7(e). Median filter operation: applied to an enhanced images by Y- direction pass through the steps above and not on an original image directly so that to reduce the number of undesired edges in original images, as a result, this will increase the accuracy of edge detection, Fig 6-7(f). Iris detection: after median filter on the image, Search for transition point (from zero to one) along the inclined scan line; this point will be considered as a boundary point between iris and sclera region, Fig 9-10(g).Iris isolation: this operation used to isolate the iris from eye image, Fig 6-7(h). a b c d e f a b c d e f a b c d e f International Journal of Computer Science and Information Security (IJCSIS), Vol. 15, No. 9, September 2017 105 https://sites.google.com/site/ijcsis/ ISSN 1947-5500
- 5. Figure. 6. Iris outer boundary localization stage in CASIA version 1.0 : (a)The original image , (b) brightness correction , (c) Applying the split operation on brightness correction image, (d) Applying the enhancement on X-direction on split operation image (e) Applying the enhancement on Y-direction on enhanced image with X- direction, (f) Applying the enhancement using median filter on enhanced image with Y-direction, (g) applying the circle on the enhancing image with median filter,(h)output of the localized iris boundary. Figure. 7. Iris outer boundary localization stage in CASIA version 4.0 interval: (a)The original image , (b) brightness correction , (c) Applying the split operation on brightness correction image, (d) Applying the enhancement on X-direction on split operation image (e) Applying the enhancement on Y-direction on enhanced image with X-direction, (f) Applying the enhancement using median filter on enhanced image with Y-direction, (g) applying the circle on the enhancing image with median filter ,(h)output of the localized iris boundary. E. Iris Normalizatoin Normalization refers to preparing a segmented iris image for the feature extraction stage. The inner and outer iris boundaries are located in an original image and can get iris region directly from an image. 1- Polar Transform using Circular distribution:- Anytime there is an opportunity for the center line of a part or component to be displaced in two directions from a target position the circular normal distribution may govern the behavior of the displacement [10]. An example of a problem involving circular normal data is shown in Fig. 8(a). The Figure shows the (x, y) positions of parts and their intended target at the origin of the (x, y) coordinate system. Individually the distributions of x and y are normal, but it is the combined effect of x and y that is of concern. The relevant displacement of a part from its intended position is measured by the radial distance r taken relative to (x, y) = (0, 0). The value of r for a part depends on its x and y displacements according to: r = (4) It is the distribution of r that is circular normal. In some cases (x, y) data may be available and in others, only the r values might be measured. A typical question to be addressed that involves the circular normal distribution for Fig.8(b), would be to determine an upper spec limit for r such that a species fraction of the observations meets the spec [11]. To calculate points along the circular boundary using polar coordinates r and 8(angle). Expressing the circle equation in parametric polar form yields the pair of equations (5) (6) When a display is generated with these equations using a fixed angular step size, a circle is plotted with equally spaced points along the circumference. Larger angular separations along the circumference can be connected with straight line segments to approximate the circular path [11]. Fig.8: (a) proposed polar transform procedure, (b) Iris region in Polar Coordinate, (c) Resize iris region. a b c d g h e f g h b c a International Journal of Computer Science and Information Security (IJCSIS), Vol. 15, No. 9, September 2017 106 https://sites.google.com/site/ijcsis/ ISSN 1947-5500
- 6. Algorithm 1:- The Proposed Polar Transform using Circular distribution Goal: will transform iris region in the original image from the Cartesian coordinates to the polar coordinates. Input: Original Image Q(x, y), radius, (x, y). Output: Iris Region in Polar Coordinate [datafeat] . Process: Step 1: Let data set is array [360, Maxradius] Step 2 : For all (i , j = 0; i < 360, j <= Maxradius - 1; i = i + 0.1,j=j+1 ) Set angle i * PI / 180; Set y xc + j * Cos(angle); Set x yc + j * Sin(angle); Get x-value = Original Image Q(x , y) Set datafeat x-value 2- Iris Region Resize The second step in iris normalization stage is iris region resize which uniform all iris region images in the system to be imaged in a fixed size. Because the iris is captured in different sizes for different people, as well as for the same person. This is due to several factors (e.g., a variation of illumination, change in distance between the camera and the eye, eye position, rotation of the camera, head tilt and rotation of the eye within the eye socket) , as shown in the Fig.8(c). IV. RESULT AND DISCUSSION The proposed iris segmentation approach tested on the iris databases (CASIA v1.0 and CASIA v4.0 interval). The eye image inputted into iris recognition system and pass through all stages as described in the following stages: Stage1: select eye image from the database mentioned above as shown figure 9, to pass through the iris boundary localization steps. Figure. 9. Original eye images: (a) CASIA v1.0, (b) CASIA v4.0-interval. Stage2: The results of pupil and iris boundaries localization steps in Figure. 10-11, on CASIA V 1.0 and CASIA V 4.0 interval respectively. Figure. 10. Pupil localization stages: (a) On CASIA version 1.0 and (b) On CASIA version 4.0 interval. Figure. 11. Iris outer boundary detection stages: (a) On CASIA version 1.0 and (b) On CASIA version 4.0 interval. Stage3: Iris normalization. The iris normalization result, transformation of iris region to Polar Coordinate in a fixed size. 1- The Iris Data Base CASIA version 1.0: as shown in Fig.12. Figure.12.Iris normalization results in CASIA V1.0 : (a) The iris after isolation, (b) Polar transform with circular distribution, (c) Iris image resize. a b a b a b a b c International Journal of Computer Science and Information Security (IJCSIS), Vol. 15, No. 9, September 2017 107 https://sites.google.com/site/ijcsis/ ISSN 1947-5500
- 7. 2- The Iris Data Base CASIA version 4.0-interval: as shown in Figure.13. Figure.13.Iris normalization results in CASIA V4.0-interval: (a) The iris after isolation, (b) Polar transform with circular distribution, (c) Iris image resize. CONCLUION In this paper, a new approach to iris normalization is presented, which focused on iris segmentation and normalization from eye image. The idea of this paper is based on finding iris boundaries of the accurate method using edge detector and circular distribution transform, this stage is very critical and effects on iris recognition system results. In addition, the proposed iris normalization stage divided into two parts iris polar transform and iris region resize to increase the accuracy and speed. Experimental results show that this approach has a good iris normalized performance. The proposed approach has implemented in c # language and Microsoft access office 2013 run under Microsoft Windows 10 Home Premium Operating System Version 6.1.7601 Service Pack 1 Build 7601. The proposed approach has an average high accuracy. Table 1: Calculation of Average Accuracy Data Base No. of Image No. Correct Segmentation No. Faulty segmentation Average Accuracy CASIA -V1.0 756 756 0 100% CASIA –V4.0 interval 1100 1100 0 100% REFERENCES [1] Bhalchandra, A., Deshpande, N., Pantawane, N. and Kharwandikar, P. (2008), ―Iris Recognition‖, Proceedings Of World Academy Of Science, Engineering And Technology, Vol. 36, pp. 1073-1078. [2] Jain, A., Flynn, P., and Ross, A. (2008), ―Handbook of Biometrics‖, Michigan State University, USA, Flynn University of Notre Dame, USA, West Virginia University, USA, pp. 79-98. [3] Nithyanandam.S , Gayathri.K.S , Priyadarshini P.L.K , "A New IRIS Normalization Process For Recognition System With Cryptographic Techniques", CSE Department, PRIST University, International Journal of Computer Science Issues, Vol. 8, Issue 4, No 1, July 2011, , Thanjavur, Tamilnadu,India. [4] J. Daugman. “How iris recognition works”, IEEE Trans. CSVT, vol. 14, no. 1, pp. 21 – 30, 2004. [5] R.P. Wildes. “Iris Recognition: An Emerging Biometric Technology”, Proceedings of the IEEE, vol. 85, pp. 1348-1363, 1997. [6] Li Ma, Y. Wang, and T. Tan. “Iris recognition using circular symmetric filters”, International Conference on Pattern Recognition, vol. 2, pp. 414-417, 2002. [7] Y. Zhu, T. Tan, and Y. Wang. “Biometric Personal Identification Based on Iris Patterns”, Proceedings of the 15th International Conference on Pattern Recognition, vol. 2, pp. 2801-2804, 2000. [8] C. Tisse, L. Martin, L. Torres, and M. Robert.. “Person identification technique using human iris recognition”, International Conference on Vision Interface. [9] W. Boles and B. Boashash, "A Human Identification Technique Using Images of the Iris and Wavelet Transform," IEEE Trans. Signal Processing, vol. 46, pp. 1185-1188, 1998. [10] http://en.wikipedia.org/wiki/Circular_distribution [11] Paul Mathews, Mathews Malnar, and Bailey, in ” The Circular Normal Distribution”, Proceedings of the IEEE 87, 2000. a b c International Journal of Computer Science and Information Security (IJCSIS), Vol. 15, No. 9, September 2017 108 https://sites.google.com/site/ijcsis/ ISSN 1947-5500