Human Recognition System based on Retina Vascular Network Characteristics

Chandrashekhar B.N Honnaraju .B Sr. Lecturer, Dept of ISE Lecturer, Dept of CSENMIT, Bangalore-64 BGS, Mandya

AbstractThis paper proposes an efficient method for Human Recognition System based on Retina Vascular Network Characteristics. Humans recognize each other according to their various characteristics for ages. We recognize others by their face when we meet them and by their voice as we speak to them. Identity verification (authentication) in computer systems has been traditionally based on something that one has (key, magnetic or chip card) or one knows (PIN, password). Things like keys or cards, however, tend to get stolen or lost and passwords are often forgotten or disclosed.Human Recognition System based on Retina Vascular Network Characteristics, This detection system can be used effectively to carry out accurate authentication of a person. Retina Vascular Network Identification Algorithm for Human Recognition. This system authorizes a person based on his retinal vascular characteristics. The system takes fundus image of the person as input, performs pre-processing and produces edge detected image. This resultant image is compared with the images stored in the database. If the image exists, then the person is authorized, else unauthorized.

Keywords-image analysis, image recognition, neural network, Image segmentation, Retina, Optic disc, Macula, Otsu method1. IntroductionThe identification procedure is based on three structural elements of the human eye retina. These are the optic nerve, the macula and the vascular network. The

reason to the selection was given by the fact that these characteristics remain unchanged through years and degradations are possible to occur only because of eye diseases, such as glaucoma and retinopathy. Human interference on the retina vascular network is not an issue at present. Part from the proposed algorithm can also be used to extract information about blood vessels network in retinal images. This information can be used to grade disease severity or as a part of automated diagnosis of diseases (Biomedical Systems).Human Recognition System based on Retina Vascular Network Characteristics, This detection system can be used effectively to carry out accurate authentication of a person. Also this detection system can be used in or suited for environments requirements requiring maximum security such as government military and banking. The iris is the coloured ring of textured tissue that surrounds the pupil of the eye. Even twins have different iris patterns and everyone’s left and right iris is different, too. Research shows that the matching accuracy of iris identification is greater than of the Retina scan is based on the blood vessel pattern in the retina of the eye. Retina scan technology is older than the iris scan technology that also uses a part of the eye. The first retinal scanning systems were launched by Eye Dentify in 1985.The retinal scanning systems are said to be very accurate. For example the EyeDentify’s retinal scanning system has reputedly never falsely verified an unauthorized user so far. The false

rejection rate, on the other side, is relatively high as it is not always easy to capture a perfect image of the retina.DNA testing.2. Literature review“Detection of Vascular Intersection in Retina Fundus Image Using Modified Cross Point Number and Neural Network Technique” By M. I. Iqbal, A. M. Aibinu, M. Nilsson, I. B. Tijani, and M. J. E. Salami.This paper talks about the application of the knowledge of digital image processing, fuzzy logic and neural network technique to detect bifurcation and vein-artery cross-over points in fundus images. The acquired images undergo pre-processing stage forIllumination equalization and noise removal.[5] Segmentation stage clusters the image into two distinct classes by the use of fuzzy c-means technique, neural network technique and modified cross-point number (MCN) methods[4] were employed for the detection of bifurcation and cross-over points. MCN uses a 5x5 window with 16 neighbouring pixels for efficient detection of bifurcation and cross over points in fundus images. Result obtained from applying this hybrid method on both real and simulated vascular points shows that this method perform better than the existing simple cross-point number (SCN) method, thus an improvement to the vascular point detection and a good tool in the monitoring and diagnosis of diabetic retinopathy.A three stage bifurcation and cross-over points detection in FI(Fundus Image) is hereby presented. These stages are: image pre processing, image segmentation and bifurcation

and cross-over point’s detection. The acquired image undergoes pre processing stage (A); for colour space conversion, illumination equalization and noise filtering using a 5x5 median filter. Image segmentation stage clusters the image into two distinct classes and detection of candidate bifurcation and cross over points is done during the third stage using the MCN and neural network technique.“Locating the Optic Nerve in a Retinal Image Using the Fuzzy Convergence of the Blood Vessels” By Adam Hoover , Michael GoldbaumIn this paper an automated method to locate the optic nerve in images of the ocular fundus has been given . Their method uses a novel algorithm we call fuzzy convergence to determine the origination of the blood vessel network. They evaluate their method using 30 images of healthy retinas and 51 images of diseased retinas, containing such diverse symptoms as tortuous vessels, choroidal neovascularization, and haemorrhages that completely obscure the actual nerve. On this difficult data set, this method achieved 89% correct detection. The optic nerve is one of the most important organs in the human retina. The central retinal artery and central retinal vein emanate through the optic nerve,[3]supplying the retina with blood. The optic nerve also serves as the conduit for the flow of information from the eye to the brain. Most retinal pathology is local in its early stages, not affecting the entire retina, so that vision impairment is more gradual. In contrast, pathology on or near the nerve can have a more severe effect in early stages, due to the necessity of the nerve for vision.Fundus Image: An image which is obtained from a fundus camera is referred to as the fundus image[2] A

fundus camera or retinal camera is a specialized low power microscope with an attached camera designed to photograph the interior surface of the eye, including the retina, optic disc, macula, and posterior pole (i.e. the fundus)A typical fundus image[2]consists of three important parts the macula, the optic nerve and the blood vessels .The optic nerve is one from which the blood vessels seems to originate from and is the brightest part of the retina.The fundus image is taken as the input for pre processing and edge detection[6] and based on which comparison is done and other operations are carried out on it.

Fig:Typical Fundus Image

2.1Image Pre-processing

Image Pre-processing essentially contains two phases. These are image enhancement and image restoration. The idea behind enhancement techniques is to bring out detail that is obscured or simply to highlight certain features of interest in an image. Image restoration techniques tend to be based on mathematical or probabilistic models of image degradation.2.1.1 Thresholding

Thresholding is defined as the process in which individual pixels in an image are marked as “object” pixels if their value is greater than some threshold value (assuming an object to be brighter than the background) and as “background” pixels otherwise.The thresholding technique that we have used is Otsu Thresholding which is an automated thresholding technique. Otsu's method is used to automatically

perform histogram shape-based image thresholding, or, the reduction of a graylevel image to a binary image. The algorithm assumes that the image to be thresholded contains two classes of pixels (e.g. foreground and background) then calculates the optimum threshold separating those two classes so that their combined spread (intra-class variance) is minimal. The extension of the original method to multi-level thresholding is referred to as the Multi Otsu method .

Fig:Before thresholding

Fig: After thresholding

2.1.2 Histogram EqualizationThis method usually increases the global contrast of many images, especially when the usable data of the image is represented by close contrast values. Through this adjustment, the intensities can be better distributed on the histogram. This allows for areas of lower local contrast to gain a higher contrast. Histogram equalization accomplishes this by effectively spreading out the most frequent intensity values.The method is useful in images with backgrounds and foregrounds that are both bright or both dark. In particular, the method can lead to better views of bone structure in x-ray images, and to better detail in photographs that are over or under-exposed. A key advantage of the method is that it is a fairly straightforward technique and an

invertible operator. So in theory, if the histogram equalization function is known, then the original histogram can be recovered. The calculation is not computationally intensive. A disadvantage of the method is that it is indiscriminate. It may increase the contrast of background noise, while decreasing the usable signal.Histogram equalization is a specific case of the more general class of histogram remapping methods. These methods seek to adjust the image to make it easier to analyze or improve visual quality

Fig:Input Image

Fig: Histogram Equalized Image

2.1.3Edge DetectionEdge detection is a fundamental tool in image processing and computer vision[6], particularly in the areas of feature detection and feature extraction, which aim at identifying points in a digital image at which the image brightness changes sharply or, more formally, has discontinuities.

The purpose of detecting sharp changes in image brightness is to capture important events and changes in properties of the world. It can be shown that under rather general assumptions for an image formation model, discontinuities in image brightness are likely to correspond to:

discontinuities in depth, discontinuities in surface

orientation, changes in material properties

and variations in scene illumination.

The edge detection techniques used here are sobel edge filtering ,prewitt edge filtering technique which are briefly described below.

2.1.3.1 Sobel FilterThe Sobel operator is used in image processing, particularly within edge detection algorithms. Technically, it is a discrete differentiation operator, computing an approximation of the gradient of the image intensity function. At each point in the image, the result of the Sobel operator is either the corresponding gradient vector or the norm of this vector. The Sobel operator is based on convolving the image with a small, separable, and integer valued filter in horizontal and vertical direction and is therefore relatively inexpensive in terms of computations. On the other hand, the gradient approximation which it produces is relatively crude, in particular for high frequency variations in the image.2.1.3.2 Prewitt FilterThe Prewitt operator is used in image processing, particularly within edge detection algorithms. Technically, it is a discrete differentiation operator, computing an approximation of the gradient of the image intensity function. At each point in the image, the result of

the Prewitt operator is either the corresponding gradient vector or the norm of this vector. The Prewitt operator is based on convolving the image with a small, separable, and integer valued filter in horizontal and vertical direction and is therefore relatively inexpensive in terms of computations. On the other hand, the gradient approximation which it produces is relatively crude, in particular for high frequency variations in the image

3 MethodologiesInitially the whole operation is divided into two phases the learning phase and the testing phase. In learning phase we try to store images of an authorized person to the database so that it can be used for comparison when he comes for authentication .The learning phase can only be carried out by the administrator so he has to login first and then only he can do further operations.The testing phase is one where an authorized or an unauthorized person comes for authentication so he first inputs his image and then comparison of the image is done with the database to check his authenticity.The main user requirements are for learning and testing phase are:-For Learning Phase:

1. The user has to login if he wants to store new images to the database through the Login option.

2. The user then has to specify the size and the name which is considered as the name of the output image that is to be stored in database using Image Size option.

3. The user then has to load the image which he wants to store in the database through the Load Image option.

4. Then the processing of the image is done through the

Process option using which the output image is obtained.

For Testing Phase:

1. The user has to load his image using the Load Image option

2. The user then has to click the compare option through which the new image is compared with images present in the database

4. Algorithm StepsThe different modules algorithms are as listed below:-

1) Image Pre-processing• Histogram

Equalization.

2) Edge detection• Sobel mask

• Prewitt mask

• Robert mask

4) Image comparison 5) Saving Image to the database4.1.1 Image Pre-processingImage pre-processing is defined as the stages before an image is processed in order to get an enhanced image through which we can get better outputs. The image pre-processing steps used here is:-1) Histogram EqualizationAlgorithm:

• Input the image• Convert to double dimensional

array• Compute Cumulative

distributive function• Compute Probability density

function• Output the image

4.1.2Edge Detection Sobel MaskAlgorithm:

• Input the image• Convert to double dimensional

array• Specify the threshold

• For each pixel in the image,• Apply Sobel mask• Compare the each resultant pixel

to threshold• If greater than threshold make

pixel black, else white• Output the image

The Sobel edge detector or the Sobel filter can be implemented in three orientations Sobel X, Sobel Y and Sobel XY. We have implemented all the three orientations and found that Sobel XY gave the best result. The output images obtained from Sobel are as follows

Fig: Implementation results of Sobel operator

Prewitt MaskAlgorithm:

• Input the image• Convert to double dimensional

array• Specify the threshold• For each pixel in the image,• Apply Prewitt mask• Compare the each resultant pixel

to threshold• If greater than threshold make

pixel black, else white• Output the image

The Prewitt edge detector or the Prewitt filter can be implemented in three orientations o Prewitt X,Prewitt Y and Prewitt XY. We have implemented allthe three orientations and found that Prewitt XY gave the best result. The output images obtained from Prewitt are as follows

Fig:Implementation results of Prewitt Operator

Robert Algorithm

• Input the image• Convert to double dimensional

array• Specify the threshold• For each pixel in the image,• Apply Robert mask• Compare the each resultant pixel

to threshold• If greater than threshold make

pixel black, else white• Output the image

Fig: Implementation results of Robert Operator

5 Image Comparisons

Learning Phase Algorithm: Input the image Convert to double dimensional

array Apply pre processing methods Apply edge detection Compare the edge detected

image with the images in the database

If the image exists then discard it and print error message

Otherwise save it in the database End

Testing Phase Algorithm: Input the image Convert to double dimensional

array Apply pre processing methods

Apply edge detection Compare the edge detected

image with the images in the database

If the image exists then print authentication success message

Else print authentication failure message

End

Saving Images in database

Algorithm:

Input the image • Convert the image into two

dimensional array• Apply pre processing and edge

detection• Then convert the obtained two

dimensional array into a raw image

• Store the raw image with the persons name obtained from input

• Then append the name to a file required for comparison

5. ResultsExperiment – A raw fundus image is given as input and then processing option is clicked which produces the following output which is compared with the database if it exists then a success message is shown or is the image does not exist in the database then a failure message message is shown

6. ConclusionIn this paper we illustrate a complete Retina Vascular Network Identification Algorithm for Human Recognition. This system authorizes a person based on his retinal vascular characteristics. The system takes fundus image of the person as input, performs pre-processing and produces edge detected image. This resultant image is compared with the images stored in the database. If the image exists, then the person is authorized, else unauthorized.We have also implemented some fundamental pre-processing techniques such as Histogram Equalization and Image thresholding to alter certain characteristics of the image. For edge detection we have used the various masks such as Robert mask, Prewitt mask and Sobel mask. These filters are valuable in detecting edges of various parts in the image.We have provided the administrator an option to add and remove the authorized person to the database. Where as a person other than administrator can only check whether he is authorized or not. Access rights to perform adding and removing a person from database is given to the administrator via username and password.

References1A. Pinz, S. Bernogger, P. Datlinger, and A. Kruger, “Mapping the human retina,” IEEE Transactions on Medical

Imaging, vol. 17, no. 4, pp. 606–619, 1998.2. A. Can, H. Shen, J. N. Turner, J. L. Tanenbaum, and B. Roysam, “Rapid automated tracing and feature extraction from retinal fundus images using direct

exploratory algorithms,” IEEE tansactions on Information Technology in Biomedicine, vol. 3,no. 2, pp. 125–137, June 1999.3. Locating the Optic Nerve in a Retinal Image Using the Fuzzy Convergence of the Blood Vessels By Adam Hoover, Michael Goldbaum

4. Detection of Vascular Intersection in Retina Fundus Image Using Modified Cross Point Number and Neural Network Technique By M. I. Iqbal, A. M. Aibinu, M. Nilsson, I. B. Tijani, and M. J. E. Salami.

5. A fuzzy impulse noise detection and reduction method chulte, S.Nachtegael, M.;   De Witte, V.;   Van der Weken, D.;   Kerre, E.E.Dept. of Appl. Math. & Comput. Sci., Ghent Univ., Gent, Belgium 

6. Local scale control for edge detection and blur estimation IEEE Trans. Pattern Anal. Mach. Intell., 20 (7) (1998), pp. 699–716

7. An improved Sobel algorithm based on median filter hunxi Ma;   Lei ang;   Wenshuo Gao;   Zhonghui Liu; Digital Media Dept., Commun. Univ. of China, Beijing, China 

8. design of an image edge detectionfilter using the sobel operator nick kanopoulos, member, ieee,nagesh vasanthavada, member, ieee,androbert baker