Abstract— The person authentication scheme based on palm print biometrics .This security system involves the techniques of pattern recognition. A biometric system is essentially a pattern recognition system that makes use of biometric traits to recognize individuals. In this paper an efficient pattern recognition technique is proposed ,the shape and contrast invariant features are extracted using local ternary pattern and the details about the illumination changes between the pixels is provided by weber local descriptor. These combined features of test image are utilized to match with original templates by using Euclidean distance for making decision on person biometric. Finally the performance of proposed algorithm will be measured with recognition accuracy and it proves that it provides better matching rate than prior approaches.
Keywords— Palm print Image, Preprocessing, LBP, LTP & Webers Local Descriptor, Similarity Measurement
I. INTRODUCTION
There are now many applications of biometrics being used or considered worldwide. Most of the applications are still at the stage of testing, and are optional for end users. Any situation that allows an interaction between man and machine is capable of incorporating biometrics. Such situations may fall into a range of application areas such as computer desktops, networks, banking, immigration, law enforcement, telecommunication networks, monitoring the time and attendance of staff. In this paper, we present our initiative work on palm print identification, which is a new attempt and necessary complement to the existing biometrics techniques. Not like hand geometry-based system that measures a hand’s size and finger length, palm print is concern with the inner surface of a hand and looks particularly at line patterns and surface shape. A palm is covered with the same kind of skin as finger tips and is larger in size than a fingertip, hence it is quite natural to think of using palm print to recognize a person, but little has been done to palm print-based personal identification. With increasing financial activities and security awareness, followed by the development of science, technology, and the progress of society, traditional authentication, such as passwords, personal identification numbers, smart cards, has been largely incapable of meeting the requirements of convenience, reliability, and security in a wide range of civilian applications. Under such circumstances, biometric identification techniques that take full advantage of intrinsic physiological and/or extrinsic behavioral characteristics of humans, such as face, iris, fingerprint, palm print, hand Shape, and handwriting, or signature, have become a powerful alternative, gaining rapid expansion. In this paper an efficient pattern recognition technique is proposed, the shape and contrast invariant features are extracted using local ternary pattern and the details about the illumination changes between the pixels is provided by weber local descriptor. This paper is organized as follows: Section II deals with the recent literatures in palm print biometrics. The proposed methodology is described in Section III. Results are discussed in Section IV. Conclusions and Future directions are given in Section V.
II. LITERATURE SURVEY
Wenxiong Kang and Qiuxia (2014) proposed a biometric system based on texture extraction and similarity measurement. The texture extraction is done based on local binary pattern. The extracted texture features are compared with the data base stored features and similarity measurement is done. The matched pixel ratio was adopted to determine the best matching region. The best matching ratio was fused with results of LBP matching to further improve the identification performance. To describe the sparse texture in palm vein images, the discrimination ability is diluted, leading to lower performance. This gives less accuracy and poor discriminatory power. [1]
Kang W, Li Y, Wu Q, Yue X proposed contact free palm vein recognition based on local invariant features. This paper presents a novel recognition approach for contact free palm vein recognition that performs feature extraction and matching on all vein texture distributed over the palm surface, including the finger veins and the palm veins, to minimize the loss of feature information. A hierarchical enhancement algorithm is used to combine DOG filter and histogram equalization, to highlight the vein texture. Root SIFT is used to overcome the projection transformation in contact free mode. Mismatching removal algorithm and LBP histogram are adopted to improve the accuracy of feature matching. [2]
Mona A. Ahmed and Hala M. Ebied (2013) presents an analysis of palm vein pattern recognition algorithms, techniques, methodologies and systems. Palm vein authentication has high level of accuracy because it is located inside the body and does not change over the life and cannot be stolen. It discusses the technical aspects of recent approaches for the following processes; detection of region of interest (ROI), segment of palm vein pattern, feature extraction, and matching. The results show that, there is no benchmark database exists for palm vein recognition. For all processes, there are many machine learning techniques with very high accuracy. [3]
Leila Mirmoham adsadeghi and Andrzej Drygajilo (2012) proposed a new approach based on local texture patterns. The local binary pattern and local derivative patterns are compared for the identification. The extracted texture features are compared with the data base stored features and identification is done. To describe the sparse texture in palm vein images, the discrimination ability is diluted, leading to lower performance. This gives less accuracy and poor discriminatory power. [4]
Ningbo Zhou and Ajay Kumar (2011) proposed two new approaches to improve the performance of palm-vein-based identification systems. This approach attempts to more effectively accommodate the potential deformations, rotational and translational changes by encoding the orientation preserving features and utilizing a novel region-based matching scheme. The palm-vein identification approach is compared with our proposed ones on two different databases that are acquired with the contactless and touch-based imaging setup. We evaluate the performance improvement in both verification and recognition scenarios. The rigorous experimental results presented in this paper, consistently conforms the superiority of the proposed approach in both the verification and recognition scenario. [5]
Mi Pan and Wenxiong Kang (2011) made a comprehensive and comparative study of three local invariant feature extraction algorithms, Scale Invariant Feature Transform (SIFT), Speeded Up Robust Features (SURF), Affine-SIFT (ASIFT) for palm vein recognition. First the images are preprocessed through histogram equalization, then the three algorithms are used to extract local features and finally results were obtained by using Euclidean distance. [6]
Hyeon Chang and Kang Ryoung (2010) proposed the finger vein recognition technique with high accuracy and rapid processing speed. A new identification method of finger vascular pattern is proposed using a weighted local binary pattern code based on support vector machine. Due to local binary pattern code features insufficient texture features are produced. This is due to shift variance. [7]
III. METHODOLOGY
Vein Pattern Analysis using Discriminative robust local ternary pattern and Weber’s local descriptor. The project presents robust palm vein recognition using hybrid texture descriptors such as discriminative robust local ternary pattern and Weber’s local descriptor for improving the recognition accuracy. In ROI Selection, entropy filter is used to extract the desired foreground region from background. Then, local threshold is used to extract the vein pattern for its texture analysis. Two textures descriptors called Weber’s local descriptors and DRLTP are proposed to extract the features about texture for recognizing with original templates.
Figure 1: Proposed methodology
DRLTP is used to provide the shape and contrast invariant features of an object. WLD provides details about illumination changes between the pixels. Euclidean distance will be used to match the features of test and original templates for making decision on person biometric.
A. Vein Pattern Detection
Palm vein pattern is extracted using image segmentation technique using local threshold algorithm. The goal of image segmentation is to cluster pixels into salient image regions, i.e., regions corresponding to individual surfaces, objects, or natural parts of objects. In computer vision segmentation refer to the process of partitioning a digital image to multiple segments. The goal of segmentation is to simplify and/or change the representation of an image into something that is more meaningful and easier to analyze Image segmentation is typically used to locate objects and boundaries (lines, curves, etc.) in images. More precisely, image segmentation is the process of assigning a label to every pixel in an image such that pixels with the same label share certain visual characteristics.
B. Weber’s Local Descriptor
In this section we give an overview of basic WLD descriptor and its extension. This descriptor represents an image as a histogram of differential excitations and gradient orientations, and has several interesting properties like robustness to noise and illumination changes, elegant detection of edges and powerful image representation. WLD descriptor is based on Weber’s Law. According to this law the ratio of the increment threshold to the background intensity is constant. Inspired by this law, Chen et.al [15] proposed WLD descriptor for texture representation. The computation of WLD descriptor involves three steps i.e. finding differential excitations, gradient orientations and building the histogram.
C. Local Ternary Pattern
LBP is sensitive to noise and small pixel value fluctuations. LTP solves this using 2thresholds to generate codes. It is more resistant to small pixel value variations and noise compared to LBP. However, it also has the same problem as LBP whereby it differentiates a bright object against a dark background and vice-versa. RLBP solves this problem for LBP by mapping a LBP code and its complement to the minimum of the two. However, RLBP cannot be applied to ULBP and L LBP of LTP. For a pair of object/background intensity inverted patterns, their ULBP codes are not complements. Similarly, their LLBP codes are also not complements. From the two LTP codes, it is observed that the 2 patterns are simply intensity inverted. However, their corresponding ULBP codes are not complements. Similarly, their corresponding LLBP codes are also not complements. The ULBP and LLBP codes are not complements. Hence, RLBP cannot be applied to ULBP and LLBP to obtain a feature that is robust to the reversal in intensity between the objects and background.
D. Euclidean Distance
Euclidean distance measures the similarity between two different feature vectors using .
Where J is the length of the feature vector, Fvi is the feature vector for individual i.
IV. RESULTS AND DISCUSSION
A set of original images with the extracted feature values are stored in the data base shown in figure 2. The objective analysis of data base image features are shown in Table 1.
Figure 2: A set of data base images
Table 1: Objective analysis of the data base image features
An input test image is given for which the vein pattern analysis is done by using local ternary pattern and weber local descriptor and then the features of test image obtained. The features obtained from the input test image are compared with the features of data base images and then similarity measurement is done. A set of input test images are shown in figure 3
Figure 3: A set of input test images
Figure 4: ROI of Input test image
Figure 5: Features of input test image by using local ternary pattern
Figure 6: Features of input test image by using weber local descriptor
Table 2: Objective analysis of the input test image features
Input Image features
contrast 5.645370732
correlation 2.187558283
Homogenity 29.17521934
Energy 6.548095741
Entropy 2.463757429
Skewness -734.7599586
Kurtosis 4646.435766
V. CONCLUSION AND FUTURE WORK
Numerous studies have utilized LTP and Weber’s law description for vein image recognition due to its higher texture representation ability. However, most studies operated on the entire image, while the fact remains that each region has a unique contribution to biometric identification via the LTP and Weber’s law algorithm. Inspired by previous studies, we proposed a matching approach in terms of partitioning of the LTP and Weber’s law histogram within the vein and its neighborhood, and comparative experiments demonstrated that our proposed method is able to highlight the texture in palm vein images and achieve better recognition performance. To improve the matching accuracy of the to-be-matched regions from contactless palm vein images, the MPC of the normalized gradient and k-means segmentation, as well as 8-neighborhood gradient module calculation methods, were utilized for texture extraction of palm vein ROI. The MPR approach was adopted for computing the matching scores 1 and the optimal matching region. Based on the optimal matching region, sub-block histogram matching was performed for the LBP code of the vein and its neighborhood, and the matching score 2 was also obtained. Finally, Similarity Measurement on Euclidean distance Measurement was utilized for score-level fusion of the two above mentioned matching methods to improve the recognition performance for contactless palm vein images.