@Article{Han04CCD,
  author  = {Robert Hanek and Michael Beetz},
  title   = {The Contracting Curve Density Algorithm: Fitting Parametric Curve Models to Images Using Local Self-adapting Separation Criteria},
  journal = {International Journal of Computer Vision},
  year    = 2004,
  volume  = {59},
  number  = {3},
  pages   = {233-258},
  bib2html_pubtype  = {Journal},
  bib2html_rescat   = {RoboCup, Vision},
  bib2html_groups   = {IAS, IU},
  bib2html_funding  = {BV},
  bib2html_keywords = {Robot, Vision},
  abstract = {The task of fitting parametric curve models to the boundaries of perceptually meaningful image
              regions is a key problem in computer vision with numerous applications, such as image segmentation,
              pose estimation, object tracking, and 3-D reconstruction. In this article, we propose the
              Contracting Curve Density (CCD) algorithm as a solution to the curve-fitting problem. The CCD
              algorithm extends the state-of-the-art in two important ways. First, it applies a novel likelihood
              function for the assessment of a fit between the curve model and the image data. This likelihood
              function can cope with highly inhomogeneous image regions, because it is formulated in terms of
              local image statistics. The local image statistics are learned on the fly from the vicinity of the
              expected curve. They provide therefore locally adapted criteria for separating the adjacent image
              regions. These local criteria replace often used predefined fixed criteria that rely on homogeneous
              image regions or specific edge properties. The second contribution is the use of blurred curve
              models as efficient means for iteratively optimizing the posterior density over possible model
              parameters. These blurred curve models enable the algorithm to trade-off two conflicting
              objectives, namely heaving a large area of convergence and achieving high accuracy. We apply the
              CCD algorithm to several challenging image segmentation and 3-D pose estimation problems. Our
              experiments with RGB images show that the CCD algorithm achieves a high level of robustness and
              subpixel accuracy even in the presence of severe texture, shading, clutter, partial occlusion, and
              strong changes of illumination.}}