New Algorithm - Simulation Connected Components Labeling for Binary Images.
Yahia Halabi, Princess Sumaya · 2013
Abstract — This research presents an algorithm for labeling connected components in binary images based on searching around black operations of an image. The proposed new algorithm walks around black to identify their boundaries. A one-dimensional array used to keep label equivalences, for uniting equivalent labels successively during the operations in different directions. The boundaries of the object are the most interesting parts to be identified. The algorithm applies two processes of labeling, one is row-wise from left to right and one is column-wise, from top to bottom at initial zero iteration. A sub image I[I,J] i+1, is defined after each labeling allocation. It is generated by subtracting the iterative labeling obtained at I [I, J] i row-wise and labeling obtained at I[I,J] i column-wise. The new image for next iteration I[I,J] i+1 = I[I,J] i- (labeling obtained from image I[I,J] i row-wise ∩ labeling obtained from image I[I,J] i