Maximally Stable Extremal Regions

   This is a MATLAB/MEX implementation of Maximally Stable Extremal
   Regions (MSER). You can:

   * Use MSER() to extract the maximally stable extremal regions from
     a given image.

   For practical coding example, see MSER_DEMO() and MSER_DEMO3().

      

  MEMBERS=ERFILL(I,ER) returns the list MEMBERS of the pixels which
  belongs to the extremal region represented by the pixel ER.
 
  The selected region is the one that contains pixel ER and of
  inensity I(ER).

  I must be of class UINT8 and ER must be a (scalar) index of the
  region representative point.

  See also MSER().

      

   R=MSER(I,DELTA) computes the Maximally Stable Extremal Regions
   (MSER) [1] of image I with stability threshold DELTA. I is any
   array of class UINT8, while DELTA is a scalar of the same class.
   R is an index set (of class UINT32) which enumerates the
   representative pixels of the detected regions.

   A region R can be recovered from a representative pixel X as the
   connected component of the level set {Y:I(Y) <= I(X)} which
   contains X.

   [R,ELL]=MSER(...) also returns ellipses fitted to the regions
   R. Each column of the matrix ELL describes an ellipse; ELL(1:D,i)
   is the ellipse center and ELL(D:end,i) the independent elments of
   the co-variance of the ellipse and D is the image domain
   dimension. The coordinates are in the same order of the indexes of
   I as an array; so the first coordinate is the row index, which in
   MATLAB convention is usally corresponds to the Y image
   coorinate. If the other convetion is needed, the elements of ELL
   must be swapped according.

   REFERENCES
   [1] J. Matas, O. Chum, M. Urban, and T. Pajdla, "Robust wide
       baseline stereo from maximally stable extremal regions," in
       Proc. BMVC, 2002.