Home

 

Members

 

Visitors

 

Projects

 

Lectures and Seminars

 

Workshops and Conferences

 

Activities

 

 

 

 

Projects

 

Title: Superresolution Imaging
Investigators: M. K. Ng, C. S. Tong, Y. M. Huang, K. T. Leung
Preiod: Jan 06 -
Description:
The recent increase in the widespread use of digital imaging technologies in consumer (e.g., digital video) and other markets (e.g., security and military) has brought with it a simultaneous demand for higher-resolution images. The demand for such high-resolution (HR) images can be partially met by algorithmic advances in super-resolution (SR) technology in additional to hardware development. Such HR images not only give the viewer a more pleasing picture but also offer additional details that are important for subsequent analysis in many applications. In this project, superresolution approach using computational, mathematical, and statistical techniques will be investigated and employed to obtain high-resolution images.

 

Title: Biometric Security and Recognition

Investigators: M. K. Ng, C. S. Tong, P. C. Yuen
Preiod: Jan 06 -
Description:
In developing real world biometric applications, the protection of biometric data (security) is one of the main concerns. Different from the password in your email account, if your account is hacked, you may use a new password. Everyone has his/her own and (almost) unique biometric data, such as fingerprint, face and iris. If one biometric data is stolen, it cannot be replaced by another biometric data. While most of the current research works focus on the recognition/verification performance of biometric authentication under different conditions and in large databases, biometric security has received less attention. The goal of this project is to study and develop methods to protect biometric data against different types of attacks while the verification/identification performance will not be degraded.

 

Title: Segmentation of Images by Active Contour
Investigators: M. K. Ng, C. S. Tong, C. P. Tam
Preiod: Jan 06 -
Description:
Active contour models are widely used in image segmentation. They evolve curves in an image, subject to constraints derived from image force and energy. Since they are using gradient information, there are some difficulties when the images corrupted by noise. Also, it is difficult to deal with textured images by applying these methods. We propose a new model integrates with region information so that it is robust to noise. Also, it can deal with textured images by combining this method together with different binarization using threshold functions.

 

 

 

 

 

 

 

 
   

Copyright © 2012 Hong Kong Baptist University. All rights reserved.

.