Biometric Fingerprint Image Enhancement,Concealment And Hashing For Content Authentication

Fingerprint image is obviously the most prominent and reliable biometric means of authentication due to its uniqueness and permanence in the entire human lifespan.If not well secured,it can however be liable to security challenges such as being hacked,counterfeited and/or repudiated through various illegal means of information falsification.Therefore,devising a means of overcoming such challenges is paramount.Concealment and hashing are the popular alternatives of watermarking for content authentication,individual identification and copyright protection of images,due to their vital properties – robustness,discrimination and security.In this thesis,we focus on 3 major information security modalities on improving,protecting and authenticating biometric data.These are enhancement,concealment and hashing.The main innovation points of this dissertation are as follows:1.A robust enhancement and centroid-based concealment scheme of fingerprint biometric data into audio signals is proposed.This scheme comes in two folds.First is to enhance the fingerprint biometric image using a minutiae recognition system by incorporating different enhancement processes.An 8-layered feature enhancement algorithm is proposed.The fingerprint image was enhanced and extracted using minutiae-based recognition system with the aim of eliminating all anomalies that comes with the image.The Effectiveness Quality Factor(EQF)and matching accuracy of the system all signifies efficiency and robustness of the enhancement scheme.And secondly,is to provide an optimised method of concealing such enhanced biometric data from being hacked or modified by intruders.As such,centroid-based audio watermarking technique is used to conceal the enhanced fingerprint biometric data into audio signals.The embedding algorithm starts by encrypting our enhanced image using chaotic logistic map prior to watermarking.It then proceeds with computing the centroid of the audio signal.Discrete Wavelet Transform(DWT)and Discrete Cosine Transform(DCT)are performed on the sub-band which carries the centroid of each audio frame,thereby embedding the encrypted watermark bits into their domain.For robustness evaluation,some signal processing operations are carried out on the watermarked signal and the outcome was intriguing as they were all counteracted.Achieved results from the performance evaluation of both contributions signify the efficiency of our proposed schemes.In essence,this contribution covers two broad areas.1)the use of image processing and recognition techniques for fingerprint image enhancement,and 2)the use of steganographic system to conceal the enhanced biometric image(security realm).2.A novel minutiae points and shape context-based hashing scheme is proposed.Fingerprint image minutiae points were extracted by incorporating their orientation and descriptors in order to generate a unique,compact and robust hash signature.This is inevitably feasible knowing that the distribution of minutiae points composes the main content structure of fingerprint images.In order to make the scheme capable of discriminating between different hash signatures and resistance to collision,fractal coding is employed for compression.This in the other hand also enable us to select any hash length of our choice,hence,ensuring convenience for its implementation.To secure the system,a secret key is used to generate a pseudorandom weight from normal distribution and embed it into the dimension of shape context random vector so they become consistent with that of the minutia descriptor.Robustness of the proposed scheme is determined by performing some content preserving attacks including noise addition,blurring and geometric distributions.Efficient results were achieved on the given attacks.Also,series of evaluations on performance comparison between the proposed and other state-of-art schemes proved our approach to be robust and secure,by yielding a better result.3.A fractal coding-based fingerprint image hashing scheme resistant to geometric attacks is proposed.This scheme addresses the issue caused due to the sole reliability of biometric authentication on minutiae in fingerprint images,where various authentication techniques involves only minutiae-wise operations which is unordered and variable in size,hence,resulting to inefficient hash generation.Therefore,in the proposed scheme,we compensated for such drawback by combining both minutiae and pixel-wise operations.The novel scheme generates a secure and robust hash using Fourier-mellin transform and fractal coding.First,Fourier-mellin transform is incorporated into the domain minutiae blocks due to its invariance property in order to improve performance under geometric operations,hence generating a fixed-length minutiae representation.Then the property of dimensionality reduction and texture compression was exploited using fractal coding in order to generate a robust and compact hash.To improve system security,encryption is performed on the extracted hash using a secret key.Experimental results demonstrated the robustness of our hashing scheme to a wide range of distortion manipulations.Additionally,performance of the proposed scheme is measured and compared with recent state-of-art techniques,and our scheme generally outperform the others by achieving a better result.