Hand Image Feature for Human Identification Samuel Adebayo Daramola a , Joke Badejo b , Isaac Samuel c , Tola Sokunbi d Department of Electrical and Information Engineering Covenant University Canaan-land Ogun State, Nigeria. a sam.daramola@covenantuniversity.edu.ng, b joke.badejo@covenantuniversity.edu.ng, c isaac.samuel@covenantuniversity.edu.ng, d tolasokunbi@gmail.com Keywords: Hand image, Centre of gravity, Hand boundary, Euclidean distance Abstract. This paper presents an algorithm for efficient personal identification using robust hand features. The feature is extracted from hand boundary points and print of hand palm. The centre of gravity of the edge map of the hand image is determined to serve as a reference point. Thereafter City block distances between the reference point and hand boundary points are found. These distance feature vectors are compared using Euclidean distance measure for effective image classification. The proposed algorithm will improve personal identification in access control and attendance record. 1. Introduction Hand is one the physical parts of human body. Human hand has many parts; these include fingers, palm, dorsal and nails. The arrangement of fingers on the palm region of hand called hand geometry is a good biometric feature that can be used for human identification. Hand geometry is one of the physiological biometric trait commonly use. The usefulness of any biometric trait depends majorly on the area of application and user acceptability [1]. Among biometric traits, hand features are very rich and easy to acquire for different applications such as access control and attendance purpose. Human identification using hand geometry is a biometric technology that involves the analysis and processing of hand image for classification process. Extraction of robust invariant features from biometric source poses serious challenges to researchers in this area of study. Many researchers have developed different feature extraction techniques for hand geometry recognition systems. In [2] twelve hand geometry features were extracted, four features are obtained using finger lengths, eight features are obtained from finger width. Also in [3] algorithm hand geometry system is proposed where three set of feature reference points from hand are established, the reference points are the tips of fingers, finger valley and centre life of the hand. Feature triangles are formed by obtaining the distances between these reference points. Furthermore [4] presented a contact free hand biometric system for real environments, geometrical features are obtained from binary images, and three independent feature vectors are formed from the index, middle and ring fingers. Each finger is characterized as a triangle. The three vertexes of the triangle are the end and the two side valleys of the finger. Also in [5] feature vector is formed by finding the Euclidean distance between hand valleys points to the major axis of each of the finger. Kullback-Leibler distance is used to compare histogram distribution between a test finger and the template one. In [6] classification of users is done using the features extracted from hand contour and palm-print region of interest. Five valley points and four peak points are detected 2013 International Conference on Advances in Information Technology Lecture Notes in Information Technology, Vol.39 978-1-61275-064-4/10/$25.00 ©2013 IERI ICAIT 2013 266