Contact Information:            
                                                                                            

• Address: Room 25204, Building 2, Area B, 74 Xuefu Road,                               
                 Heilongjiang University, Harbin, China, 150080
    
• Phone:    +86 13654662298
    
• Email:     lijuanxiao2298@hotmail.com

Research Interests:

• Image/Video processing, with particular interests in object tracking 
                                             
• Development of parallel image processing algorithms based on GPU with CUDA

Education:

• M.S. 2006-present: Computer Application Technology, School of Computer Science and Technology, Heilongjiang University,                                 Harbin, China.
  
  Main Courses:         Digital Image Processing, Combinatorics, Parallel Computing, etc.
       
• B.S. 2002-2006:      Information and Computing Secience (Applied Mathematics), PLA Information Engineering University,                                 Zhengzhou, China.
  
  Main Courses:         Mathematical Analysis, Advanced Algebra, Discrete Mathematics, Modern Algebra, Numerical  Analysis,                                 Probability and Statistics, C Programming Language,  C++ Programming Language, etc.

Publications:

• Xiao Lijuan and Li Peihua. Improvement on Mean Shift based Tracking Using Secong-Order Information. In 19th International Conference on Pattern Recognition (ICPR 2008), December 2008, Accepted for Publication.

• Li Peihua and Xiao Lijuan. Histogram-Based Partial Differential Equation for Object Tracking. In 7th International Conference on Advances in Pattern Recognition (ICAPR 2009), February 2009, Accepted for Publication.
  

• Li Peihua, Xiao Lijuan. Mean Shift Based Object Tracking: Reconsidered and Extension. Submitted for reveiw, 2009.

• Li Peihua, Liu Xiaomin, Xiao Lijuan and Qi Song. Robust and Accurate Iris Segmentation in Highly Noisy Iris Images. Image and Vision Computing Journal, Submitted for Review, 2009.

Current Research:

Parallel Algorithm Based on GPU:

• Parallel Mean Shift based tracking algorithm is proposed by using off-the-shell Graphics Processing Unit (GPU). The algorithm consists in representing object color efficiently with a small number of bins based on K-Means clustering, together with a sequence of parallel algorithms called kernels executed on GPU.

Tracking Objects with Similarity or Affine Transformation:

• Traditional Mean Shift tracking algorithm has unsatisfactory performance in locating objects that undergo similarity of affine transformation, because the object shape is invariably modeled by an upright ellipse despite pose variation. By developing the candidate model that incorporates similarity or affine transformation, we derive efficient optimization algorithms to handle complex pose change. The proposed algorithms are effective and also enjoy real time implementation.

Histogram-Based Partial Differential Equation for Object Tracking :

• By using two different techniques of shape derivative and variational derivative, we derive the partial differential equation (PDE) that describes the evolution of the object contour. Level set algorithm is used to compute the solution of the PDE.

Iris Processing :

• Skin classifier as a pre-processing step in iris segmentation

Use color histogram and Gaussian mixture to represent skin and non-skin distribution in iris images, respectively. The skin classifier is derived via likelihood ratio and the pixel is regarded as skin if the ratio is bigger than the prescribed threshold. The threshold can be adjusted to compromise between the false positive rate and the correct detection rate.

Computer Skills:

• C, C++:      Proficient in C/C++
• CUDA:       Good at developement of parallel algorithm with CUDA
• LATEX:      My favourite typsetting tool
• MATLAB:   Proficient in MATLAB

Honors:

• Take part in Noisy Iris Chanllenge Evaluation (NICE.I) and be Ranked #4 among all 97 registered teams from 35 countries.

Language Skills: