SOM based Face Recognition using Steganography and DWT Compression Techniques Sujatha B M 1 , Nayina Ramapur 2 , Shubhangi Lagali 2 , K Suresh Babu 3 , K B Raja 3 , Venugopal K R 3 1 Acharya Institute of Technology, Bangalore, India. 2 Saitektronix Private Limited, Bangalore, India 3 University Visvesvaraya College of Engineering, Bangalore University, Bangalore, India. 1 sujathabm2005@gmail.com, 2 nayana.rni@gmail.com Abstract- Biometrics is used in day to day life of human beings to access mobile phones, computers, vehicles etc. In this paper, we propose SOM based Face Recognition using Steganography and DWT Compression Techniques. The various available standard face databases such as ORL, JAFFE, NRI, YALE, Indian male and Indian female are used to test the performance of the algorithm. The number of face images per person is reduced using stenography compression technique. The preprocessing techniques such as resize and Gaussian filter are applied on reduced number of face images for uniform size and good quality of images. The initial features are extracted using Discrete Wavelet Transform (DWT) and Local Binary Pattern (LBP). The DWT and LBP features are fused using arithmetic additions and connected to input of Self Organizing Map (SOM). The final features are extracted from SOM. The test image features are compared with face database images using Euclidian Distance (ED) to compute performance parameters. It is observed that the performance of the proposed method is better than the existing methods. I. INTRODUCTION Biometrics is the metrics which is related to the characteristics of humans. Biometric verification is the means through which a person can be uniquely identified by evaluating some of the distinguishing biological traits. Unique identifiers which are commonly used are fingerprints, earlobe geometry, hand geometry, retina and iris patterns, voice, DNA, and signatures. The law enforcement uses face recognition system to identify an individual from an array of people and it has been quite reliable. A facial recognition system is a software application based on biometrics that is capable of identifying a person from a digital image or a video frame. The identity of an individual is verified by analyzing the selected facial features from an image with a facial database and is mainly used for security purposes. Most recent facial recognition systems works on numeric codes known as face prints that are identified on a human face. Face points are used to analyze the characteristics of an individual’s face; these characteristics can be classified into nose length or width, eye socket depth and cheekbone size. Given the required prerequisites, face recognition system can effectively identify an individual with the help of face prints. If the target’s face is obscured or inclined rather than facing forward, or if there is no sufficient International Journal of Computer Science and Information Security (IJCSIS), Vol. 14, No. 9, September 2016 806 https://sites.google.com/site/ijcsis/ ISSN 1947-5500