Non Uniform Multiresolution Method for Optical Flow and Phase Portrait Models: Environmental Applications

详细信息    查看全文

• 作者：Cohen ; Isaac ; Herlin ; Isabelle
• 关键词：non uniform multiresolution ; optical flow ; non quadratic regularization ; finite element method ; adaptive mesh ; phase portrait ; flow pattern classification ; ocean and atmospheric circulation
• 刊名：International Journal of Computer Vision
• 出版年：1999
• 出版时间：1999
• 年：1999
• 卷：33
• 期：1
• 页码：29-49
• 全文大小：929 KB
• 刊物类别：Computer Science
• 刊物主题：Computer Imaging, Vision, Pattern Recognition and Graphics
  Artificial Intelligence and Robotics
  Image Processing and Computer Vision
  Pattern Recognition
  
• 出版者：Springer Netherlands
• ISSN：1573-1405

文摘

In this paper we define a complete framework for processing largeimage sequences for a global monitoring of short range oceanographicand atmospheric processes. This framework is basedon the use of a non quadratic regularization technique for optical flowcomputation that preserves flow discontinuities. We also show thatusing an appropriate tessellation of the image according to anestimate of the motion field can improve optical flow accuracy andyields more reliable flows. This method defines a non uniformmultiresolution approach for coarse to fine gridgeneration. It allows to locally increase the resolution of the gridaccording to the studied problem. Each added node refines the grid ina region of interest and increases the numerical accuracy of thesolution in this region. We make use of such a method forsolving the optical flow equation with a non quadratic regularizationscheme allowing the computation of optical flow field while preservingits discontinuities. The second part of the paper deals with theinterpretation of the obtained displacement field. For this purpose a phase portrait model used along with a new formulation of the approximation ofan oriented flow field allowing to consider arbitrary polynomial phaseportrait models for characterizing salient flow features. This newframework is used for processing oceanographic and atmospheric imagesequences and presents an alternative to complex physical modelingtechniques.