An efficient scan algorithm for block-based connected component labeling
Wan Yu Chang, Chung-Cheng Chiu · 2014
In this paper, an improved block-based connected-component labeling algorithm that enhances labeling speed is proposed. By analyzing the neighboring operations in a block-based scan mask, the proposed algorithm successfully decreases the scan mask from the original 20 pixels to ten pixels. The simplifying neighborhood operations also produce the decision tables of the new scan mask for the block-based connected-component labeling algorithm. Furthermore, the proposed algorithm efficiently integrates the new block-based scan mask into the original scan through two procedures for different situations in order to enhance performance. The experimental results show that the proposed algorithm can acquire better performance than other connected-component labeling algorithms on various resolution images.