Open Thesis Topics

1. Scene learning and object recognition

Scope: Digital image processing, machine learning

Level: PhD/Masters

Description: Feature detection/extraction from the information readily available in the sequence of images is particularly important to identify objects in the images. This is immensely helpful for building autonomous systems and intelligent surveillance systems. The purpose of this project is two-fold: (i) the student is required to extract robust features from the images and analyze their performance, and (ii) implementing (supervised/unsupervised) machine learning based approach to identify and classify the objects based on the extracted features’ database. The first part of the project can be implemented at Masters Level which can later be extended to PhD thesis for robust features extraction and objects classification. Alternatively, the whole project can be selected as a PhD research topic.

Skills required: C/C++, OpenCv, matlab.

 Examples of scene learning and object classification

Examples of scene learning and object classification

2. Person Re-Identification

Scope: Image processing, machine learning, intelligence surveillance
Level: PhD/Masters

Description: The goal of this project is to develop an interactive visual search method that finds a given pedestrian in a large archive of other camera views efficiently. A user-selected pedestrian image or sequence is used to obtain initial discriminative features and an initial ranked list of hypothetical matches. A discriminative pedestrian recognition model is learned in an on-line manner by user interaction assigning positive and negative labels to the initially retrieved results and on-line boosting for feature selection. This enables that the best discriminative features for the current query are selected.

Skills required: C/C++, MATLAB, statistics, etc. 

Person re-identification from image database of a camera network

3. Fast image enhancement using CUDA (Complete Unified Device Architecture)

Scope: Digital image processing, parallel processing

Level: PhD/Masters

Description: Image enhancement has huge applications in robotics, autonomous systems, night vision and medical imaging. A number of techniques have been proposed in the literature to enhance the quality of images which can be broadly classified into spatial domain methods (operating directly on pixels), and frequency domain methods (operating on the Fourier transform of an image). However, most of the image enhancement methods are computation intensive and are not suitable for real-time image processing. In this project, student will be required to go through the existing techniques of image enhancement and select a suitable approach to implement on CUDA which is a parallel computing platform and programming model invented by NVIDIA. It enables dramatic increases in computing performance by harnessing the power of the graphics processing unit (GPU). The project can further be extended to real-time video quality enhancement.

Skills required: C/C++ or MATLAB, CUDA.

4. Object detection, segmentation and classification using random forest 

Scope: Image processing, machine learning
Level: PhD/Masters

Description: Object detection in large-scale real-world scenes requires efficient multi-class detection approaches. Random forests have been shown to handle large training datasets and many classes for object detection efficiently. The aim of this thesis is to compare the existing object classification techniques, addressing their issues, optimizing performance, and proposing a generic framework of random forests for multi-class object detection in images. 

Skills required: C/C++, MATLAB, statistics, etc.

Example of vehicle detection and segmentation in images

5. Ant colony optimization based, resource aware routing in WSN 

Scope: Wireless Sensor Networks, simulation & modeling, optimization

Level: PhD/Masters

Description: Resource aware routing in Wireless Sensor Networks (WSNs) is a challenging issue. Not only efficient routing deals with a number of factors including the shortest path between sender and receiver, reliable delivery of the messages, satisfactory coverage of the events, etc, but also consider the resources of WSN nodes. One of the resources worthy of major concern in a WSN is the limited energy source. This creates a constrained optimization problem where the objective is the efficient routing of the information/messages with a number of constraints including the energy level of a node, processing capabilities and the area of coverage. This project is related to a thorough study of ant colony optimization which is one of the potential candidates to solve the above mentioned optimization problem. The student in this project is then required to implement the ant colony optimization algorithm on a simple WSN network for resource aware routing.

Skills required: MATLAB, C/C++

Example of routing in WSN using ant colony optimization

6. Network banner grabbing for validating and implementing security 

Scope: Networking & communication

Level: Masters/Undergraduate

Description: Banner Grabbing is an enumeration technique used to glean information about computer systems on a network and the services running its open ports. Administrators can use this to take inventory of the systems and services on their network. An intruder however can use banner grabbing in order to find network hosts that are running versions of applications and operating systems with known exploits. In this Master’s thesis, students will develop an application (first for a workstation and then possibly for a Smartphone) to grab banners of the connected devices (including PCs, smart phones, tablets, etc) and analyzing the banners to validate and implement security over the network and informing the clients (via a message) about the level of security (low, medium, high) they are having.

Skills required: basics of networking protocols.


Example of network banner grabbing

7. Network traffic prediction 

Scope: Networking & communication, simulation & modelling.

Level: Masters/Undergraduate

Description: Unexpected message peaks create challenges in all networks and service delivery systems due to the limited processing capacity. In a load balancing environment (e.g., cloud), it is particularly important to predict the incoming workload to provide a satisfactory level of service.  In real time situations it is, however, impractical to rely on simple reactive resource provisioning models because by the time new resources are allocated, the existing capacity would not be able to meet the increased workload, resulting in an increase in message queuing time. Instead, the system should be capable to predict the load changes in advance. In the thesis you should investigate different prediction models and implement the chosen models in a proof of concept networking environment. 

Skills required: MATLAB, basics of statistics and networking protocols.


Example network traffic with respect to time

8. Development of a Cluster Based Cloud with Hadoop Framework 

Scope: Software Engineering, parallel and distributed computing

Level: Masters/Undergraduate

Description: Cloud computing is likely to become the largest distributed computing paradigm in future. Not only it helps reducing the investment cost for the clients machines, but also provide interactive access to the clients’ data with a high level of location transparency. Apart from the storage and application services provided by a cloud server to the clients, a cloud also provides services for processing computationally-intensive and time-consuming tasks which would otherwise require huge computational resources. In this thesis, the students are required to develop a cloud for the QUEST University which will be based on a cluster capable of providing parallel computing and storage services to its clients. The parallel and fast computing services will be based on the open-source Apache Hadoop framework using the computational paradigm MapReduce.  Using this cloud, the clients should be able to (i) access their applications and data from anywhere at any time, and (ii) migrating the computation-intensive tasks to the cloud.

Skills required: basics of networking, C/C++


Example cluster cloud

9. Intelligent web search using semantics 

Scope: software engineering, machine learning, data and web mining

Level: Masters/Undergraduate

Description: Semantic search seeks to improve search accuracy by understanding searcher intent and the contextual meaning of terms. This entails using experts' knowledge to make search results more refined and ultimately improve the result over conventional methods. For example, an enhanced search engine could use Wikipedia articles and or terminologies such as WordNet to extract important phrases /words which are closely related to the user's search request. This design-and-build project would require the student to identify such useful semantic sources for search and develop an application to retrieve and present results which are closely linked with the user's search. For instance, if "The Beatles" is searched, many pages talking about the Beatles (the group and not insects) will be shown, but the semantic search engine could show about Beatles records, films, books, etc.

Skills required: basics of data mining, python/C, etc.


Example of normal and intelligent web search based on semantics

10. Developing a software tool for analyzing business data from social networks 

Scope: Data and web mining

Level: Masters/Undergraduate

Description: Services like Twitter and Facebook provide users with an opportunity to post real-time updates and status reports. Many businesses are now using these tools to provide client services ranging from advertising to customer care. The aim of this design-and-build project is to provide access to a variety of data mining techniques in order to allow businesses to better understand real-time data as it evolves. For example, the marketing department of a large firm may wish to understand how users are responding to a recent advertising initiative or marketing campaign. The student will develop a web application to provide the end user with access to a set of analysis and visualization tools that are capable of extracting valuable information from the real-time stream. For example, the project will look at text analysis techniques to extract frequent terms, location tags, shared URLs etc. In addition, special effort will be devoted to the development of a visual dashboard that is capable of presenting real-time data in a meaningful and compelling manner and in a way that the end user can readily understand.

Skills required: basics of data mining, python/C, etc.


Example of twitter chatting trend analysis

11. The risks and benefits of cloud computing – managing costs and security concerns 

Scope: Distributed computing, security.

Level: Masters/Undergraduate

Description: The widespread popularity of cloud computing is readily understandable as it has much to offer in terms of low cost overheads, flexibility and modularity. However the widespread use of cloud computing also raises a number of security concerns especially if users are utilizing unsecured networks or engaging in lax security practices such as sharing passwords and failing to log out of networks properly.  Initial evidence suggests that many of the security concerns relate to poor user habits as opposed to weaknesses in the hardware and software itself. Accordingly this work examines the extent of the risks and exposure this presents in order to assess how these risks can be overcome.

Skills required: Statistical methods


Example of cloud computing parameters worth analysis

12. Vehicle Tracking System 

Scope: Embedded systems, microprocessor/microcontroller interfacing, electronics

Level: Undergraduate

Description: This project presents an automotive localization system using GPS and GSM-SMS services. The system permits localization of the automobile and transmitting the position to the owner on his mobile phone as a short message (SMS) at his request. The system can be interconnected with the car alarm system and alert the owner on his mobile phone. This tracking system is composed of a GPS receiver, a microcontroller and a GSM Modem. GPS receiver gets the location information from satellites in the form of latitude and longitude. The microcontroller processes this information and this processed information is sent to the user/owner using GSM modem.  The presented application is a low cost solution for automobile position and status and very useful in case of car theft situations. The proposed solution can be used in other types of application, where the information needed is requested rarely and at irregular period of time (when requested). The system can be further improved by designing a more interactive web-based application at the client side.

Skills required: basics of embedded systems and electronics, C/C++.

Depiction of the vehicle tracking system

13. Network modeling, simulation and analysis using OPNET 

Scope: Networking & communication, simulation & modelling.

Level: Masters/Undergraduate

Description: In the network research area, it is very costly to deploy a complete test bed containing multiple networked computers, routers and data links to validate and verify a certain network protocol or a specific network algorithm. The network simulators in these circumstances save a lot of money and time in accomplishing this task. Network simulators are also particularly useful in allowing the network designers to test new networking protocols or to change the existing protocols in a controlled and reproducible manner. In this Master’s thesis, student will perform a comprehensive survey on current network simulators, their main features, consider their advantages and disadvantages, and discuss the current and future developments. The student will then model and simulate a chosen (university’s) network and analyze the realistic simulated network to compare the impact of different technology designs.

Skills required: basics of networking protocols, OPNET.

Example of network simulation using OPNET

14. Solving Shortest-Path Problem Using Any Colony Optimization

Scope: Optimization, WSN routing

Level: Masters/Undergraduate

The Shortest Path Problem (SPP) can be stated as follows: given a weighted network G in which each edge has a weight representing the length of the edge, the objective of Shortest Path Problem is to find a shortest path from some node s, called the source to another node, called the sink, in G. The SPP problem is the basic ingredient problem in network optimization problems. It also has numerous applications in other areas.
Many traditional algorithms have been proposed for the SPP problem. However, in nature the ant colony can also solve the SPP problem when they are searching for food. From this researchers developed Ant Colony Optimisation (ACO) algorithms. The aim of this project is to design, implement, and demonstrate ACO algorithms and some tranditional optimization algorithms for SPPs with comparison of their performance in either C++ or Java.

Shortest-path finding with any colony optimization

15. Motion Based Multiple Object Tracking

Scope: Digital image processing, surveillance

Level: Undergraduate

This project is intended to perform automatic detection and motion-based tracking of moving objects in a video from a stationary camera. Detection of moving objects and motion-based tracking are important components of many computer vision applications, including activity recognition, traffic monitoring, and automotive safety. 

Motion-based multiple object tracking

16. Vehicle Counting & Classification

Scope: Digital image processing, surveillance

Level: Masters/undergraduate

Congested Roads and Traffic signals are becoming more and more common due to increasing number of vehicles. This requires effective planning and management and for that requires efficient vehicle monitoring and counting methods. Also the vehicles has to be classified as a two wheeler occupies a lesser space than a truck. This image processing project takes images from traffic junction cameras and performs a real time analysis to do this. Using this information a lot of processes can be automated from traffic signals to toll gates.

Real-time detection, counting and classification of vehicles

17. Automatically Detect and Recognize Text in Natural Images

Scope: Digital image processing

Level: Masters/Undergraduate

This project investigates how to detect regions containing text in an image. It is a common task performed on unstructured scenes, for example when capturing video from a moving vehicle for the purpose of alerting a driver about a road sign. Segmenting out the text from a cluttered scene greatly helps with additional tasks such as optical character recognition (OCR).

Detection of text in images

18. Performance Comparison of Image Features for Object Detection

Scope: Digital image processing

Level: Masters/Undergraduate

In this project, students will be required to compare the quality and accuracy of different types of image features used for object detection and recognition, e.g., BRISK, FAST, HARRIS, MinEigen, MSERF, SIRFT, SURF, HAAR and HOG, etc. The quality of these features will be compared for different images and scenarios. The comparative study will suggest what feature performs best for a given image. The task of feature comparison will be carried out in MATLAB. 

Object detection with HOG features

19. Image Category Classification
Scope: Digital image processing

Level: Masters/Undergraduate

This project is aimed at using a 'bag of features' approach for image category classification. This technique is also often referred to as bag of words. Visual image categorization is a process of assigning a category label to an image under test. Categories may contain images representing just about anything, for example, dogs, cats, trains, boats.

Image category classification using bag of features

20. Distance Calculation From a Stereo Video
Scope: Digital image processing

Level: Masters/Undergraduate

This project is aimed at  detecting people in video taken with a calibrated stereo camera and determining their distances from the camera. Besides people, the project can be targeted to detect other objects and their distances.

People detection and distance calculation from stereo video

21. Comparison of HAAR, HOG & LBP Features For Object Detection
Scope: Digital image processing

Level: Masters/Undergraduate

Object detection has immense significance in computer vision. A number of object detectors exists which are trained by using different types of image/object features. The most popular types of image features used for object detection include HAAR, Histogram of Oriented Gradients (HOG), and Local Binary Pattern (LBP) features. This project aims at comparing the relative performance of these image features and their detectors by training these detectors for a certain object detection (e.g., people, vehicle, etc). 

Detection of different types of objects using object features

22. Image Forgery Detection
Scope: Digital image processing

Level: Masters

The rapid growth of image editing softwares has given rise to large amounts of doctored images circulating in our daily lives, generating a great demand for automatic forgery detection algorithms in order to determine the authenticity of a candidate image in a timely fashion. A good forgery detection algorithm should be passive and blind, requiring no extra prior knowledge of the image content or any embedded watermarks. By analyzing the abnormal behaviors of doctored images from authentic images, one can design forgery detectors based on a collection of cues in the image formation process. The goal of this student project is to perform a literature overview on the topic of image forgery detection for forensic applications. Further, an overview on probably existing freeware and commercial software should be provided, as well as an exemplary algorithm should be implemented and evaluated.

Image forgery detection

23. Suspicious Object Detection in Public Places

Scope: Digital image processing

Level: Masters

The automatic detection of objects that are abandoned in public places such as railway stations, airports, bus stands or hospitals where there is large movement of people is an important task. This helps highly in video surveillance, especially in the fight against terrorism and crime, and public safety. This cannot be done using any ground sensors or even with policing because of vast majority of people and area involved. This project is aimed to detect such objects from a sequence of images captured by a camera.

Suspicious/abandoned object detection

24. Image Encryption & Decryption For Secure Communication

Scope: Digital image processing

Level: Masters/PhD

Information security has always been ensured with data encryption and authentication techniques. Nowadays a lot of generic data encryption and decryption algorithm has been developed. The secrecy of communication is maintained by secret key exchange. In effect the strength of the algorithm depends solely on the length of the key. The presented work aims at secure image transmission using randomness in encryption algorithm, thereby creating more confusion to obtain the original data. Since the encryption process is one way function, the artificial neural networks are best suited for this purpose.

Image encryption and decryption

25. Dimension Measurement of Buildings Using Image Processing

Scope: Digital image processing

Level: Masters

Imagine measuring a toll apartment without going near or using measurement tape. It’s possible using a real time camera and image processing techniques. Many times measuring buildings or large industrial products or machines are complex because of its huge size or just because of a hazardous environment that is present around it. In such cases a camera along with a low cost hardware can give an accurate and more reliable dimension measurements such as height and width of any material.

Dimension calculation using image processing

26. Intensity-Based Registration of Medical MRI Images

Scope: Digital image processing

Level: Masters/undergraduate

Image registration is the process of transforming different sets of data into one coordinate system. Data may be multiple photographs, data from different sensors, times, depths, or viewpoints. It is used in computer vision, medical imaging, military automatic target recognition, and compiling and analyzing images and data from satellitesThe work on this project is focused on aligning magnetic resonance images (MRI) to a common coordinate system using intensity-based image registration. Unlike some other techniques, it does not find features or use control points. Intensity-based registration is often well-suited for medical and remotely sensed imagery.

Unregistered and registered images

27. Comparative Analysis of Machine Learning Techniques

Scope: Machine Learning

Level: Masters

This research project is aimed at comparing the performance of several machine learning techniques on a selected dataset. The performance comparison will be carried out for a number of machine learning techniques including neural networks, logistic regression, discriminant analysis, K-nearest neighbors, naive Bayes, support vector machines, decision trees, and tree bagger etc. 

SelectionFile type iconFile nameDescriptionSizeRevisionTimeUser