Postgraduate level Projects
Project - Towards Better Performance: Development of an Automated Tool for Performance Evaluation in Sports using Trajectory Analysis.
Dr.U.A.J. Pinidiyaarachchi - Supervisor
Prof. S.R. Kodituwakku – Co-supervisor
Mr. K.D.G. Kuruppu – MPhil Candidate
The primary concern of this work is on weightlifting. It is very important to correct the techniques in weightlifting to be recognized in ranking. The expected outcome of this research is a system that is developed to assist athletes, in their training. Hence the system will allow athletes to correct their techniques very easily without having to spend large amount of money. This research is focused on developing a system with following capabilities: Any athlete can obtain data on his trajectory very easily and correct his techniques. They can obtain other data like bar travel distance, total time for lift, maximum height, max velocity etc. They can also compare ideal weightlifter trajectory with one’s own trajectory to get an estimate on ranking.
Funding Source:National Research Council
Project:- Object detection in a complex image and retrieval of images stored in meta image databases
Dr.U.A.J. Pinidiyaarachchi - Supervisor
Dr. S. Mahesan – Co-supervisor;University of Jaffna
Ms. B. Ganesharajah – MPhil Candidate
Content-based retrieval of images is a challenging field in processing visual information as it requires perceptual abilities in computational form. The intelligent retrieval of images from large databases is the focus of substantial research efforts within the computer vision community. Identification of objects in image is one of the fundamental problems in computer vision. Several existing image retrieval systems do not provide better generalized performance and they are time consuming. The objective of this work is the development of novel method to retrieve specific objects from a complex image by querying the properties of these images. We also aim to reduce the workload of human operators such as image analysts.
Project -Monitoring System for ATM transactions and Performance Analysis of the System in a parallel environment
MPhil Student : P. Arumainayagam
Supervisors: E.Y.A Charles and S. R. Kodituwakku
Data Stream Management System is one of the most popular research areas in computer science. The most challenging fact in managing data stream is handling the data which change with time. Using Sequential Query Language (SQL), can easily query over finite datasets (DBMS). But in data streams data and its quantity are change with time. So querying over these streams is difficult and challenging.
Data Stream Management System (DSMS) is an important research area in Data Base Management Systems (DBMS). Querying over finite set of dataset is simple with SQL. But in DSMS data and its density will change with time. So dataset also change time to time. Accessing this data and querying over this data is very difficult and a challenging task. Nowadays ATM card usage is increasing broadly and quickly in the overall world as well as fraud usage of this cards also increase. Therefore, nowadays ATM fraud detection is also a very important issue to be addressed. This project mainly focuses on ATM fraud detection and analyzing the performance of this system in a parallel environment. Not only concern with fraud detection it also concern on most of its operations.
Main aim of this research is processing ATM transactions and finding the fraud transactions. This can be useful for the people who use these ATM cards nowadays. Most of the people use these cards and these fraud activities are also happening in many places. This research will be mainly focus on preventing the fraudulent use of ATM cards. Also find the performance of the system in a parallel environment.
Project:- Performance Comparison and Enhancement of Content Based Image Retrieval Techniques
MPhil Student: S. Selvarajah
Supervisor: S. R. Kodituwakku
Content-based image retrieval (CBIR) is becoming more and more important with the advance of multimedia and imaging technology. Among many retrieval features associated with CBIR, color, texture and shape are the most widely used features. The aim of this research is to evaluate the performance of different feature descriptors proposed for CBIR and to investigate enhanced feature descriptor with better performance.
Since different descriptors based on color, texture and shape features have been developed and have been used, this project aims to analyses such features and their performances. Then the performances of individual descriptors are compared to determine which of them are more effective in retrieving images. Finally the performances of combined individual features are analyzed in order to enhance the retrieval performance.
Combination of colour and texture feature descriptors, combination of colour features, and combination of colour and shape feature descriptors have already been investigated. Experimental evaluation is carried out using publicly available databases.