Related essay: Authentication by the hand vein structure
Biometrics are automated methods of recognizing a person based on a physiological or behavioural characteristics. There are number of physiological and behavioural characteristics . Physiological characteristics are like iris, vein technology, face, palm print and behavioural characteristics such as gait recognition , voice, odour etc. as the frauds are increasing day by day we need more secure methods to increase the security and biometrics are of the best approach. There are number of applications such as airports, banks, hospital security and access to restricted areas.
The security field uses three different types of authentication:
- What you know – a password, PIN, or piece of personal information
- What you have – a card key, smart card, or token
- What you are – a biometric.
Of these, a biometric is the most secure authentication tool which can’t be borrowed, stolen, or forgotten, and forging one is practically impossible.
The personal identification using hand and palm vein has gained more and more research attentions these years. There are number of properties of Palm Vein:1) the vein information is unique to each even between twins; 2) it is difficult to be damaged; 3) it is highly secured as resides insides the body. Because of these, hand and palm vein seems a better biometric feature that finger print and face.
The Palm Vein Recognition extracts the features if the palm and match it against the database. The palm has the broader area and contain more complex vascular pattern which contain the large number of features which is used for recognition. The vein pattern does not change over time as it is present inside the body. This feature makes it suitable for one-to-many matching, for which fingerprint and face recognition may not be robust.
A Palm Vein characteristic has gained more interest in authentication because of its uniqueness even between twins. The most important advantage of palm vein is that it is exist only for live humans.
In the palm vein recognition system, the vascular patterns of an individual’s palm is personal identification data, as a palm is a broader area than finger and a complicated vein pattern and thus contains a wealth of different features for personal identification. The palm is an ideal part of the body for this technology; it normally does not contain hair as back of the hand which can create obstacle for photographing the vein pattern. It is less susceptible to a change in skin colour, unlike a finger or the back of a hand.
The paper published by Jing-Wein Wang , et.al on ‘Building Palm Vein Capturing System for Extraction' in which they create a vein capturing system which contain three phases. First phase is pre-processing in which the various steps are carried out to improve the image quality such as a high-boot filter for image enhancement, a bi-level thresholding, a median filter to eliminate noise in the image. Second phase is feature extraction in which the vein features are extracted by ‘pixel by pixel’ scanning. Third step is post-processing in which thinning is done to remove the unnecessary information
A paper published by Huan Zhang, et.al, on ‘A Palm Vein Recognition System' In which they purposed a capturing device to capture the image. They selected CCD camera which have low response in the wavelength of visible light. In pre-processing the inscribed circle-based segmentation which extracts the ROI from the original palm vein image. The Gaussian smooth filter, local contrast enhancement is used to enhance the ROI image. Then the vein patterns are extracted which include vein length and minutiae for recognition and matching. At last, the thinning method is used to thin and repair the vein line.
A paper published by Sahar Bayoumi, et.al, on ‘PCA-based Palm Vein Authentication System' In which they proposed a system for authentication using Palm Vein based on using principle component analysis (PCA) for feature extraction. Palm vein images of dorsa are captured by by infrared camera by which the database is creted. PCA is applied to generate vector of features of first and second order correlation pixel. A matching process is done to recognise the person. Experiments show that there system is able to recognize human with accuracy 85%.
A paper published by Yi-Bo Zhang , et. al on’ Palm Vein Extraction and Matching for Personal Authentication' in which they represent a system of personal authentication using palm vein which includes four phases. In first phase palm image is captured in which they use low cost CCD camera to capture the infrared palm images. Second phase is detection of Region of Interest in which a small area (128*128 pixels) of a palm image is located as ROI to extract the features and to compare different palms. Third phase is palm vein extraction by multiscale filtering to improve the performance of vein detection. Fourth phase is matching.
A paper published by Qing Rao , et,al on ‘Personal Identification For Single Sample Using Finger Vein Location and Direction coding' in which they propose a structured personal identification approach using finger vein Location and Direction Coding(LDC). First of all, they design a device with near-infrared (NIR) light source for imaging, with the help of which the database will be created. Then they do pre-processing by using size and brightness normalization. They normalize the size by linear interpolation. Furthermore, finger vein LDC is proposed and performed, which creates a structured feature image for each finger vein. Then local threshold method is used to further segmentation of the ambiguous pixels by finding mean and deviation of neighbouring pixel then median filter is used to smoothen the image by removing the noise and at last vein location and direction coding is done to yield the vein feature map with direction information. Finally, the structured feature image is utilized to conduct the personal identification
A paper published by Debnath Bhattacharyya, et al; ‘Vascular Pattern Analysis towards Pervasive Palm Vein Authentication' in which they proposethree different algorithm for the image preprocessing which will smooth the image. These three different processes are: a. Vascular Pattern Marker Algorithm (VPMA); in which the two pass masking is used such as horizontal and vertical kernels to smoothen the image b. Vascular Pattern Extractor Algorithm (VPEA); in which Thresholding is done. Vascular Pattern Thinning Algorithm (VPTA) in which thinning is done for capturing the Vascular Pattern of hand Palm of an individual.
A paper published by Gin-Der Wu, et.al, on ‘An enhanced discriminability recurrent fuzzy neural network for temporal classification problems’ in which they proposes an enhanced discriminability recurrent fuzzy neural network for temporal classification problems. To consider classification problems, the most important consideration is the ‘discriminability’. To enhance the ‘discriminability’, the feedback topology of the proposed fuzzy neural network Analysis results indicate that the proposed fuzzy neural network exhibits excellent classification performance.
3. PROBLEM FORMULATION
After studying the research papers I came to know about the overall description of Palm Vein Authentication and its corresponding method. The lots of work has been done on the Palm Vein technology but there is still a scope of further improvement. The accuracy and response time of the existing system is slow which can be improved. Error extractions due to bad quality of the palm vein pattern images may lead to the fatal errors of the process. The quality of images are low. Recognition rate of the existing system is not good enough to be a real system.
Palm recognition can be used as an extended biometric print technique as an alternative to Finger vein recognition. Because Palm is of larger area than finger, it is more accurate comparatively. Moreover, palm vein can overcome the finger injury problem. Fingers are most prone to injury. Hand Palm is the area with less chances of injury.
1. Error extractions due to bad quality of the palm vein pattern images may lead to the fatal errors of the process .
2. The accuracy and response time of the existing system is slow .
3. The images are of poor quality [3-1].
4. Recognition rate of the existing system is not good enough to be a real system .
5. Threshold value is assumed which is not always a criteria .
1. Pre-processing of Palm Vein image and ROI extraction .
2. Features will be extracted and Recognition will be done by Fuzzy-Neuro Technique.
3. Performance evaluation / Comparison with existing technique.
Proposed palm vein based authentication algorithm will be designed to be more accurate and fault tolerant. This proposed model will be developed to perform various important tasks. The purposed system is operated in two modes: training mode and testing mode.
Training Set Generation
In this all image samples are first pre-processed to remove the noise from the image then the image pre-processed are used to find the Region of Interest in which the specified area is specified from which the value of features are extracted and these values are stored for the later purposes. Testing mode contain the no of phases which is described below one by one with explanation:
a. Pre-Processing and ROI Extraction:
Pre-processing is done by median filter and ROI will be extracted by the following algorithm:
a) Binaries the input image.
b) Calculate the centeroid of the binary image.
c) After this draw square around the centeroid by the distance from the centeroid of the image.
d) Extract the sub image within the ROI.
b. Features Extraction:
First the segmentation is done by region based segmentation in which the four neighbour concept is used. In this the seed pixel i.e the starting pixel is found and then their 4 neighbouring pixel is found then the distance is evaluated from the seed pixel to the neighbouring pixel . the shortest distance pixel is further evaluated using same procedure until all pixels are processed. Then the features will be extracted by finding the curvature points
For matching the Fuzzy-Neuro technique will be used as follows:
I. Features of trained image and testing image will be matched by XOR operation.
II. Accordingly, some Fuzzy rules were generated
III. These fuzzy rules will act like the atoms to the neural network.
IV. We will multiply the values of these atoms and found whether it will reach the threshold value.
V. If the criteria is matched according to the neural network then the testing image is authenticated otherwise not.
‘ Chip: Intel Pentium(Pentium IV or above)/ intel Celeron/Intel Xeon/ Intel core/AMD Athlon/ AMD opteron
‘ RAM: 500 MB
‘ Hard Disk: 512MB/(1024 MB recommended)
‘ 16-, 24- or 32-bit OpenGLcapable graphics adapter
‘ DirectX 9.0 or later
‘ Operating system: Window XP (Service Pack 1 or 2) / window 2000 / windows server 2003/ Window Vista, Windows7/8
‘ MATLAB version 7.2
‘ Image Processing Toolbox