Home › Research › Research Groups
Multiple Description Coding for diversity systems
Multiple description coding (MDC) and reconstruction of images has been received considerable attention in the signal processing community for the last decade because of its interesting and excellent features over unreliable communication networks such as internet, ATM networks, packet-switched networks, wireless communication networks over fading multipath channel and so on. The aim of this research is to develop an efficient and reliable MDC scheme over unreliable communication networks. A general framework of multiple description robust communication system is targeted with block-based dc separation method. The advantage of this system is that, if all the channels work, a high quality, possibly lossless, reconstruction can be achieved from all the received descriptions. On the other hand, a lower but still acceptable quality can be achieved if some of the channels do not work. Our target is to analyze 2, 4 and 8 channels cases in order to obtain an excellent result for transmitting images in a diversity system.
Facial Expressions Recognition
Automatic facial expression recognition is a challenging research issue with usefulness in a variety of fields. It is playing increasingly important role in the fields of human computer interaction, data-driven animation, criminal identification, psychological treatment etc. Success of most facial image analysis solutions depend on an effective facial feature representation. The objective of this research is the recognition of facial expression using a new appearance-based facial feature called the Local Monotonic Pattern (LMP). LMP can extract robust facial feature from a face image that gives accurate and reliable recognition performance for expression recognition. The basic idea is to apply LMP operator on a pixel, finds the monotonic intensity transition of neighboring pixels at different radii. The final feature vector is expected to be a collation of these histograms. This feature vector will then employ to classify expressions with well known machine learning method: Support Vector Machine (SVM).
Multi-Resolution Analysis of Ultra Wide Band Impulse Radio Signal
Wavelet Packet Transform (WPT) based multi-resolution analysis technique is suitable for Ultra Wide Band (UWB) Impulse Radio (IR) signal detection from a noisy environment. This research is going on WPT based signal detection technique based on calculation of energies of the coefficients of each WPT terminal-node and by using an improved threshold calculation technique.
Dr. Abdur Razzak
Md. Asheque Imran