Nilanjan Ray

Professor
Department of Computing Science
University of Alberta, Canada


Document Binarization

Document thresholding is an important step in intelligent document analysis. For hand-written documents an interesting cue is the width of pen strokes that more or less remain constant. We utilized this shape characteristic to recursively subdivide an image and apply separate intensity threshold in each subdivision.

Related Publications
  • J. Shi, N. Ray and H Zhang, “Shape Based Local Thresholding for Binarization of Document Images,” Pattern Recognition Letters, vol.33, pp.24-32, 2012.

Variational Image Thresholding

We apply variational optimization principle to compute a threshold surface for an image. The threshold surface passes through image edges.

Related Publications
  • B. Saha and N. Ray, “Image thresholding by variational minimax optimization,” Pattern Recognition, vol.42, no.5, pp.843-856, May 2009.
  • N. Ray and B. Saha, “Edge sensitive variational image thresholding,” Proceedings of IEEE ICIP, 2007, pp.VI - 37 - VI –40, 2007.