Semantic-Based CBIR (Content-Based Image Retrieval)
Faculty In charge:  Dr. Qi

The advances in multimedia content have produced an enormous number of digital imagery archives in a variety of application domains. All this imagery information is useful only when one can access it efficiently. Thus, the demands for an intelligent CBIR system, capable of effective understanding and accurate retrieval of visual information, are growing worldwide. The goal of this project is to explore important enabling techniques for an image retrieval system to lay a solid foundation for future multimedia search engines. We will focus on developing an automatic image annotation component, a relevance feedback-based image retrieval component, or a semantic-correlation-based image retrieval component.

Image Forensic and Authentication
Faculty In charge:  Dr. Qi

Trustworthy photographs play an important role in many applications such as news reporting, intelligence information gathering, criminal investigation, security surveillance, and health care. However, the trustworthiness of pictures could no longer be taken or granted with the advent of digital age. The goal of this project is to develop novel forensic and authentication methods for protecting image copyright and detecting image forgeries. These techniques will significantly improve the current status of DRM (Digital Rights Management). We will focus on developing robust watermarking schemes for achieving resilience to geometric distortions. We will also develop a completely blind and passive system for detecting digital photograph tampering without using extra encryption, extracting digital signatures, and embedding copyright information.

Object Detection (Fire and Flame Detection, Barcode Reader, Human Detection/Face Detection,
Car Detection/License Plate Recognition, etc.)
Faculty In charge:  Dr. Qi

Fire and flames are one of the leading hazards affecting everyday life around the world. It is also known that fires and burns are the second most common cause of death to children under 10 in U.S., second only to automobile crashes. Conventional point smoke and fire detectors are widely used in buildings and cannot be used in open spaces and large covered areas. As a result, video based fire/flame detection systems can be useful for detecting fire and flame in large auditoriums, tunnels, atriums, etc. Such systems can be used to process the outputs of CCTV (Closed Circuit Television) surveillance cameras for possessing fire detection capability. The goal of this project is to develop a few novel and real-time CVIP techniques to detect fire and flame in large and open spaces.

There are currently 11.4 million visually impaired individuals living in the US. Grocery shopping is an activity that presents a functional barrier to independence for many visually impaired people. The goal of this project is to develop a reliable computer vision-based barcode reader algorithm to help the visually impaired persons to shop independently.

Project 4:

Cooperative Network Monitoring
Faculty In charge:  Dr. Mano

As a network grows in size and complexity, the difficulty of monitoring network traffic for malicious behavior increases. In the meantime, defending against new network attack models requires a more fine-grained approach to monitoring (i.e., switch-level packet payload analysis), which makes network traffic analysis an even more resource intensive process. The goal of this project is to investigate how a diverse set of heterogeneous network devices can coordinate traffic analysis results to identify malicious network activities. This cooperative approach to network monitoring will enable security practitioners to even identify collaborative insider attacks that utilize covert inter-switch communication.

*Possible Projects:  Students may design their own Computer Vision and Multi-Agent related projects.