SCALP: Superpixels with Contour Adherence using Linear Path

Iterative clustering framework

• SCALP is based on the Simple Linear Iterative Clustering (SLIC) framework [4]
• Initial grid partition into S×S superpixels, and clustering of pixels within 2S×2S boxes for each superpixel S_k
• Color and spatial distances are considered to gather spatially close and homogeneous pixels so
  the clustering distance between a pixel p and a superpixel S_k is:     D(p,S_k) = d_color(p,S_k) + d_spatial(p,S_k)

       

Use of pixel neighborhood

• SCALP considers the neighborhood information to produce a more accurate decomposition and be robust to noise
• The pixel neighborhood N(p) improves the color distance:     D(p,S_k) = d_color(N(p),S_k) + d_spatial(p,S_k)

Linear path definition

• Linear path Pk
  p={q} between the position X_p of a pixel p and the barycenter X_k of the considered superpixel S_k

Color distance on linear path

• SCALP considers the color distance on Pk
  p:     D(p,S_k) = (d_color(N(p),S_k) + d_color(Pk
  p,S_k)) + d_spatial(p,S_k)
• The color distance on the linear path prevents irregular shapes to appear:

Contour distance on linear path

• A contour prior map can be given to SCALP to enforce the respect of image objects.
• The maximum of contour intensity on Pk
  p is considered in a new term d_contour(Pk
  p)

• SCALP clustering distance is defined as:   D(p,S_k) =

  (

  d_color(N(p),S_k) + d_color(Pk
  p,S_k) + d_spatial(p,S_k)

  )

  d_contour(Pk
  p)