Research

Multispectral Palmprint Recognition

This research investigates the use of multispectral images for palmprint recognition. A novel representation and hash table encoding, termed as 'Contour Code', is proposed for multispectral palmprint recognition. A reliable technique for region of interest (ROI) extraction from palm images, acquired with non-contact sensors, is designed. The Contour Code is binarized into an efficient hash table structure that only requires indexing and summation operations for simultaneous one-to-many matching with an embedded score level fusion of multiple bands. Comprehensive experiments are performed on two multispectral palmprint databases. [Publications][Code]

Image Steganography and Steganalysis

In this project, a universal steganalysis technique for the detection of hidden messages in digital images is developed. Steganography is being used to hide information in images and later transfer them through the internet without any suspicion. Steganalysis aims at detecting the presence of hidden content in images by using statistical approach. As modern steganographic systems are able to preserve the statistics of the original image while embedding a message, the task of steganalysis becomes more challenging. Messages are hidden in a database containing 1338 images for each steganography algorithm at varying embedding rates. Steganalysis techniques are then developed using Discrete Wavelet Transform, Discrete Cosine Transform, Discrete Contourlet Transform and their combinations to extract features for steganalysis. Experiments performed on various databases of test images reveal the effectiveness of the proposed steganalysis approach by demonstrating above 93% detection.