Research Article
Salient Region Detection by Fusing Foreground and Background Cues Extracted from Single Image
Algorithm 1
Label propagation via boundary nodes.
Input: | : Row-normalized affinity matrix | : Set of given boundary nodes’ labels | : Set of unlabelled nodes | | Initialize, set for and for | while VAR > threshold do | for do | | end for | | = var() | end while | BG = normalize() | BG() = map(BG) | Output: | Regional background probability weight map BG() generate by background labels |
|