Home
People
Projects
Publications
Links
 

CGIP Home > Projects

  
Visual Style Algorithmic Properties of Visual Style
Philip Buchanan
This research focuses on the development of measures for stylistic classification, replication and generation of graphics. Such metrics will be based on contextural geometric units as well as image properties.

Publications:
 >  CGI-2012
 >  IVCNZ-2012
  
Crowd Simulation Crowd Simulation and Animation
Bo Li
This research aims to develop efficient collision detection and motion planning algorithms for crowd simulation and animation. An extended oriented bounding volume is used for geometry and motion representation.

Publications:
 >  IVCNZ-2012
  
Lung Segmentation HRCT Lung Segmentation
Huaqing Chen
A fully automatic method is developed for identifying and segmenting lung regions in High-Resolution Computed Tomography (HRCT) images.

Publications:
 >  IVCNZ-2011
  
Tissue Tissue Image Processing
Jacky Tay
This project looks at the relationship between various multifractal measures of cell structures in tissues and the corresponding pleomorphic scores assigned by pathologists. The Holder exponent of local intensity variations is used to identify Mitotic cells

Publications:
 >  IVCNZ-2011
  
Facial Expression Facial Expression Representation
Mohammad Obaid
Quadratic deformation models are applied to motion capture data to obtain generic transformation parameters that represent different types of facial expressions.  These  parameters can then be used to transfer the expression to two or three dimensional models.

Publications:
 >  CGIV-2009       SIGGRAPH-2010
 >  ACE-2009        CGIV-2010
 >  DICTA-2009
  
Multifractals Tissue Image Classification Using Multifractals
Anna Hemsley
The theoretical and implementation aspects for multifractal based classification techniques are studied and used for representing certain shape characteristics of tissue and cell images using multifractal spectra.
Project funded by:
 >  ESI Bright Ideas (2008),   CoE Operational Research Grant (2009)

Publications:
 >  IJMDEM-2010
 >  ISM-2009
  
Facial Animation Facial Expression Animation
Jacky Tay
Muscle regions are automatically mapped to a two-dimensional sketch based model or an image, based on primary feature points, and then deformed using rubber-sheet transformation coefficients derived for a given facial expression. The model is animated by interpolating the transformation coefficients from the neutral expression to the given expression for each muscle region. Additional facial features are added for enhancing the realism of the generated facial animation sequence.
 
Publications:
 >  IVCNZ-2009
  
J2ME J2ME Computer Vision Library for Mobile Devices
Jian Gu
This research explores the feasibility of using Java (J2ME) to develop cross platform computer vision applications for mobile phones with integrated cameras. Several AR applications on J2ME enabled mobile phones are developed and tested using the library. Experiments are conducted to evaluate the compatibility, portability and efficiency of vision algorithms and  functions implemented in the library.

Publications:
 >  IVCNZ-2008
  
Painterly Rendering Fast Algorithm for Painterly Rendering on Mobile Devices
Charles Han
A low-complexity painterly rendering algorithm suitable for implementation on mobile devices is developed. A sequential connected component labelling algorithm is used to represent regions of similar color in an index buffer and a color table. All subsequent processing is done on the index buffer to minimise computational complexity.

Publications:
 >  TPCG-2008
  
Pattern Recognition Radial-Tchebichef Invariants for Pattern Recognition
R Mukundan
A new class of global shape descriptors based on Tchebichef polynomials is developed for applications in pattern recognition.  The definition of Radial-Tchebichef moments is very similar to Zernike moments, except that the kernel functions are discrete orthogonal polynomials. Rotational invariants can be easily derived from these moment functions because of their radial-polar structure.

Publications:
 >  BMVC-2009
  
Geometric Moments Geometric Moments Using Schlick's Approximation
R Mukundan
A modified version of normalized geometric moments can be created by using Schlick's approximation for the exponential terms. Schlick’s approximation is useful in significantly reducing the computational time required for evaluating the moment kernel. The modified moments have a better dynamic range of values, particularly for high orders.

Publications:
 >  IJPRAI-2008
 >  ISVC-2007

Web site and all contents © CGIP 2011-12.