Registration of images capture by different sensors is difficult because of the dissimilarities in the imaging process, the geometric errors introduced, and the specific type of information recorded in the images. An example is the registration of RGB color images and a set of multi-spectral images for the same scene. This project will study this topic by developing scale and rotation invariant key point extraction algorithms, and then using graph-based method for robust image registration.
Understand image registration process and advance the knowledge in this topic.
Background in computer vision, pattern recognition or machine learning. Experiences with C/C++ and/or Matlab.
Linda G. Shapiro and George C. Stockman, "Computer Vision", Prentice Hall, 2001. David A. Forsyth, Jean Ponce, "Computer Vision: A Modern Approach", Prentice Hall, 2003.
Research experiences in the Australian National University.