Linear Variation and Optimized Algorithm of Connected-Component Labeling in Binary Image

Linear variation is a topological characteristic of a function of two variables. The problem of linear variation computing can be reduced to the problem of counting connected components in a binary image with eight-connected connectivity. The proposed method is essentially a modification of some known raster algorithms for connected components labeling that groups pixels into 2× 2 cells. A performance benchmark of the proposed method applied to some approaches on random images is provided.