Research

Performance Analysis of High Speed Low Power DWT Architectures for Image Fusion/ Registration/ Compression algorithms for Micro Air Vehicle Applications.

By : Venkateshappa Guide : Dr. Cyril Prasanna Raj P.

Home land security, border security, remote sensing, surveillance and tracking are some of the primary application areas where micro air vehicles play a vital role. Due to geographical challenges, rough terrains and unpredictable environments and climatic conditions humans beings are finding it difficult to monitor and provide security across the border as well as within home land. Use of micro air vehicles are the only possible solutions in near future.

Cameras that are mounted on micro air vehicles reduces payload and eliminates multiple sensors. However images/video sequences captured occupy huge bandwidth and hence need to be compressed and transmitted to base station for processing. As MAVs need to operate in day light and night conditions, IR cameras and thermal cameras along with visible light camera is used to acquire data. The acquired data is compressed and transmitted to the base station, at the base station the image registration and fusion is performed to extract information for monitoring and control.

Design, modeling and VLSI implementation of low power high speed wavelet based 3D architectures for image registration.

By : Sunitha P. H. Guide : Dr. Sreerama Reddy G. M.

With the growth in VLSI technology leading to nanometer transistors 3D VLSI IC technology is gaining momentum for high speed and low power signal and image processing applications. Medical imaging and aerial imaging are the major applications where 3D imaging plays an important role with image registration. In this research work, 3D architectures that are parallel, pipelined and systolic architectures are analyzed for their performances. Image registration algorithms that are used for 3D images are also analyzed for their performances, suitable algorithms for image registration based on Wavelet Transform approaches will be identified and implemented using 3D architectures optimizing for area, power and speed performances.

Image registration helps physicians monitor the progress of cancerous tissue. Better registration techniques could give physicians new and more effective ways of monitoring and treating the millions of patients suffering from cancer. Also, image registration would be useful for more precise targeting of radiation therapy against tumors. Better image registration could also lead to improvements in image-guided surgery – a surgical procedure where the surgeon uses indirect visualization to operate. . Turning this data into information requires image registration to be accurate enough to consistently compare images across time and modalities.

Design, Modeling and Performance Analysis of Novel Algorithms for EEG Based Brain Computer Interface for Paralytic Patients.

By : Tejaswini C. Guide : Dr. Sreerama Reddy G. M.

Paralytic patients are also called as “locked-in” patients. Because of a stroke, traumatic brain injury, cerebral palsy or a degenerative neurological disease such as amyotrophic lateral sclerosis, their entire motor system is paralyzed. These patients are fully conscious and alert, but are unable to use their muscles; they are therefore unable to communicate their needs, wishes and emotions. In these patients a healthy brain is actually locked into a paralyzed body. There is no cure, and the cause of amyotrophic lateral sclerosis, one of the major causes of locked-in syndrome, is still unknown.

This research aims to investigate intelligent systems which facilitate development of assistive robotic device for people with severe movement disability such as in locked-in patients. The project objective investigates a brain-computer interface (BCI) that allows a paralytic patient person to control robotic devices by thought alone. The research primarily involves developing advanced signal processing algorithms to extract information from BCI tasks-related electroencephalogram (EEG), commonly known as brain-waves, and requires investigations into signal pre-processing, feature extraction and classification aspects using innovative statistical methods and computational intelligence approaches.

VLSI Implementation of Low Power MB-OFDM PHY Base band Modem for Wireless Sensor Network.

By : Naveen H. Guide : Dr. Sreerama Reddy G. M.

Efficient design and implementation of wireless sensor networks has become a hot area of research in recent years, due to the vast potential of sensor networks to enable applications that connect the physical world to the virtual world. WSNs are usually composed of small, low cost devices that communicate wirelessly and have the capabilities of processing, sensing and storing By networking large numbers of tiny sensor nodes, it is possible to obtain data about physical phenomena that was difficult or impossible to obtain in more conventional ways. In the coming years, as advances in micro-fabrication technology allow the cost of manufacturing sensor nodes to continue to drop, increasing deployments of wireless sensor networks are expected, with the networks eventually growing to large numbers of nodes.

Wireless sensor networks are emerging as one of the most challenging technologies to meet the communication requirements and data monitoring requirements in several time critical applications. Recently ITU has proposed new set of protocols for wireless sensor networks and road map indicates that in another few years there will be high data rate and data transfer over wireless sensor networks. Software implementation and validation have been successfully accomplished by many of the researchers for demonstration of WSN capabilities. However, for real time implementation and practical purpose there is a need for hardware platforms. In this work, two major research activities are taken up:

1. Design and modeling of MB-OFDM UWB physical layer for high data rate and development of novel algorithms for hardware implementation
2. Design and implementation of novel architectures for the proposed MB-OFDM UWB on hardware platforms

Design, Modeling and FPGA Implementation of Robust Noise Filtering Algorithms for Under Water Imaging Applications.

By : Azra Jeelani Guide : Dr. Veena M. B.

A total 3, 90,884 accidental deaths were reported in the country during the year 2011. A total of 6, 94,390 cases of Un-Natural Accidents have caused 3, 67,194 deaths and rendered 5, 06,348 people injured during 2011. The major un-natural causes of accidental deaths were Road Accidents (37.3%), Railway Accidents and Rail-Road Accidents (7.6%), Poisoning (8.0%), Drowning (8.1%), Sudden Deaths (7.3%) and Fire Accidents (6.7%). 29708 deaths have occurred during the year 2011 due to drowning (boat capsize). Recovering the bodies from the deep water is a herculean task, as trained divers are required to dive deep into the water, identify the body, and carefully remove the body from the bottom. Currently there are very limited solutions to detect the body without diving into the waterfront. Divers need to swim underneath the water and with trial and error locates the bodies and recover them.

In this research work, we propose a through water imaging equipment that can scan the water front in defined areas and generates images of the water bed indicating the location of drowned bodies. In order to achieve higher performance and reliability an array of sonar signals are transmitted and received. The received signals are processed to generate images, using which human bodies are detected. The received signals need to be preprocessed before information retrieval could happen. Noise plays an important role in distorting the underwater images. One of the major source of noise is the speckle noise, that need to be filtered from sonar images.

Digital UWB Through Wall Imaging for Human Body Motion Detection on VLSI Platform.

By : Vinod Kumar B. L. Guide : Dr. Cyril Prasanna Raj P.

Through-the-Wall radar imaging (TWRI) is emerging as a viable technology for providing high quality imagery of enclosed structures. The ability to locate moving targets inside a building with a sensor situated at standoff range outside the building would greatly improve situational awareness on the urban battlefield.

Through wall imaging technique is required for time critical applications such as disaster management, rescue operations and terrorist tracking behind walls. Micro air vehicles are found to be used as mobile surveillance unit to track terrorists hiding. Detection of human motion behind walls needs to be carried out within few microseconds or milliseconds. TWI techniques are mostly software oriented and hence there is a need for SoC that can process data and retrieve information in real time. In this work, novel algorithms will be developed to identify human motion behind walls and suitable architectures designed for high speed data processing. In the proposed research work, the architecture shown in Figure 2 is modified for human body detection in motion using digital UWB. The modified algorithm will be implemented on FPGA for time critical applications.

EEG Based Emotion Detection and Classification Using Multiwavelets and Neural Networks.

By : Mangalagowri S. G. Guide : Dr. Cyril Prasanna Raj P.

The field of Brain-Computer Interfaces (BCIs) has gained enormous popularity during the last few years. The Brain Computer Interface (BCI) systems enable people suffering from severe neuromuscular disorders to operate devices and applications through their mental activities . Bio-signals can be acquired from an organ or a nervous system. The Brain Computer Interface (BCI) system involves recording and analyzing the Electro-Encephalogram signals from brain using electrodes.

Monitoring of the user's brain activity results in brain signals (e.g. electric or hemodynamic brain activity indicators), which are processed to obtain features that can be grouped into a feature vector. The latter is translated into a command to execute an action on the BCI application (e.g. wheelchair, cursor on the screen, spelling device).The result of such an action can be perceived by the user who can modulate her/his brain activity to accomplish her/his intents.

It is identified that there is a need for novel algorithms for emotion classification that can achieve more than 70% accuracy. Also it is required to classify emotions in abnormal patients those who suffer from neurological disorders. In this proposed research work, novel algorithms that combine neural network approaches and frequency domain approaches will be developed for emotion classification in abnormal patients.

VLSI Implementation of Brain Computer Interface.

By : Azarathamma S. Guide : Dr. Cyril Prasanna Raj P.

The field of Brain-Computer Interfaces (BCIs) has gained enormous popularity during the last few years. The Brain Computer Interface (BCI) systems enable people suffering from severe neuromuscular disorders to operate devices and applications through their mental activities . Bio-signals can be acquired from an organ or a nervous system. The Brain Computer Interface (BCI) system involves recording and analyzing the Electro-Encephalogram signals from brain using electrodes.

Monitoring of the user's brain activity results in brain signals (e.g. electric or hemodynamic brain activity indicators), which are processed to obtain features that can be grouped into a feature vector. The latter is translated into a command to execute an action on the BCI application (e.g. wheelchair, cursor on the screen, spelling device).The result of such an action can be perceived by the user who can modulate her/his brain activity to accomplish her/his intents.

It is identified that there is a need for novel algorithms for emotion classification that can achieve more than 70% accuracy. Also it is required to classify emotions in abnormal patients those who suffer from neurological disorders. In this proposed research work, novel algorithms that combine neural network approaches and frequency domain approaches will be developed for emotion classification in abnormal patients.

Design and Analysis of Bioinspired Motion Detection and Velocity Estimation Algorithm and Architecture for Autonomous Navigation of Mavs.

Highly accurate real-time stabilization and navigation of UGVs, MAVs, humanoids and vehicles is a major research focus of robotics and automation. Research on autonomous navigation and guidance is influenced by the anatomy of insect’s eyes. A fly’s panoramic vision system comprises at its front end several thousand photoreceptors feeding into a 2D array of motion detecting neurons which the animal uses for dynamic visuo-motor pose and gaze stabilization and navigation. Insects such as fly possess a visual system that is fast, precise and reliable hence, they are able to execute flawless landing on the rim of a container, instantaneously avoid obstacles while flying.

Insects have immobile eyes with fixed-focus optics, the eyes are also positioned very close hence possess inferior spatial acuity. The behavior of flying insects is dominated by visual control and hence insects use visual feedback to stabilize flight, control speed and measure self motion. Highly accurate real-time stabilization and navigation of UGVs of MAVs is major research interest as these systems are used for surveillance, security, search and rescue mission. Thus design, modeling and implementation of Bioinspired visual system on UGVs and MAVs could be an accessible methodology to replace the traditional image processing algorithms in controlling flying robots.