Research
Research interests, projects and publications
The main focus of my current research is Image and Video Processing, Cryptography and Security and the appication of Neural networks to solve problems within these domains. More specifically, I have concentrated my attention to the topics of:
• Object segmentation in complex videos
• Cryptographic systems based on Hopfield neural networks
• Lattice basis reduction techniques
• Design of secure hash-functions for biometics based authentication
The core of my Ph.D. Dissertation work is a neural network approach to segmentation of foreground objects from videos that contain background objects underoging complex movement of and illumination changes. The Background modelling Neural Network (BNN) is a neural network designed to model the background in such videos and classify the pixels as those pertinent to background or foreground objects.
Hopfield neural networks as nonlinear associative-memory systems with stochastic error have been used in several cryptosystems. The cryptoanalysis of such applications is a part of the research I engaged in with my friend and colleague Daniel Socek.
The lattice basis reduction problem is related to a number of interesting problems. My interest in it is to try and come up with an algorithm that would perform beter than the famous LLL technique and thus form a viable attack to the NTRU cryptosystem.
More recently, I am looking at the possibility of employing neural networks to achieve efficient secure hashing functionality required by the biometrics-based authentication systems.