LAB PEOPLE

 

12.jpgLei Guan (M.S. Student)


Office: EA 723B

 

Phone: (215) 204-3160

 

E-mail: lei.guan@temple.edu

 

Lei Guan received the B.S. degree in Measuring & Control Technology & Instrumentations from Nanjing University of Science and Technology, Jiangsu, China, in 2008. He is currently pursuing the M.S. degree at Temple University. His research interests include image registration, image fusion, face recognition and pattern recognition.

  Research interests and specialties

■    Image Registration

■    Image Fusion

■    Face Recognition

■    Pattern Recognition

Current research topics

■  Registration and Fusion of Visible and Thermal IR Images for Face Recognition

*    Motivation

Despite a significant level of maturity of face recognition with visible imagery in controlled situations, face recognition is still a challenging task in uncontrolled environment, such as in illumination variant conditions, with different facial expressions, and different poses. Thermal IR imagery, which can capture the emitted energy from an object, is more robust to illumination variations and thus offers a promising alternative to visible imagery for face recognition. However, thermal IR imagery is sensitive to changes in body and ambient temperature and opaque to glass. In contrast to thermal IR imagery, visible imagery is almost robust to these factors but more sensitive to illumination variations. Considering visible and thermal IR imagery can provide complementary information, registration and fusion of them provide a viable way to improve the overall performance of face recognition under uncontrolled situations.

 

*    Objective

The objective of registration and fusion of visible and thermal IR images is to stably use information from both spectrums to improve the performance of face recognition under uncontrolled situations like variant illumination conditions, partially occluded situations.

 

*    Proposed method

(1) Registration of Visible and Thermal IR Images:

Registration of Visible and Thermal IR images

(2) Fusion of Visible and Thermal IR Images:

      Wavelet Transform and Particle Swarm Optimization

 

*    Preliminary Results

(1) Partially Occluded Face:

                    

(2) Lateral Illumination Condition:

                   

 

 

Courses

Digital Signal Processing: A Computer-Based Approach

Applications of Digital Signal Processing

Probability and Random Processing

Machine Learning for Robot Perception