• Shape statistics based on warp: Spine x-rays Normally shape variation is described as linear independent statistics on the position of the landmarks describing the shape. This is a very simple model. Novel methodologies instead warp the image so that the shape in the concrete image become identical to the prototype shape. Then statistics on the warps may be performed instead leading to better statistical models that may to a higher degree with fewer parameters describe the natural shape variation. A project in this involves implementing a warp method, warp a number of spine x-rays, make linear statistics on the warps, and test the influence on the statistical shape model.
  
  
• Detection of aortic calcification in x-rays Aortic calcifications are visible as small areas looking like a shadow on the image. In this project a method for detecting these is implemented and the performance is tested. You will design a bank of filters for describing the calcification and use this in a K nearest neighbour classifier for detecting calcifications in new x-rays.
  
  
• Detection of fragments in human eggs After insemination cell division take place in the human egg. In this process also a number of cell fragments appear. The amount of fragments is considered an indicator for high risk of abortion. In this project you will design a fragment classifier and test it on a number of images. Methodology is close to the one of aortic clacification detection.
  
  
• Restoration of movies Methodologies have been developed for restoring old movies includes dust and scratches. A major problem however is to detect which areas are actually damages. In this project you will design a scratch detector, and test its performance. The necessary ingridients are optic flow computation and scale-space filters.

Please notice! You sign up for this project cluster as if it was a normal course. But please remark:

• enrolment in a project cluster is not considered a binding course registration (only relevant to MSc students)
  
• You can register for a project cluster as a fourth activity (besides of three courses).
  
• You/your group have to register your project in the Project Base before the deadline for 12 week projects, 16 week projects, 4 week projects or theses/final projects. Please find the deadlines here http://www1.itu.dk/sw923.asp
  
• The project and its title you register in
  the project base, will appear on the diploma,
  when you have passed the project. (Please be
  aware that the project cluster by itself is not
  registered on the diploma).
• Single-subject students at Open University and guest students from other universities interested in the project cluster, please contact the Student Administration Office, phone +45 72 18 52 05.