A shrinking algorithm for binary images to preserve topology

The two main factors that determine the quality of images from scaling algorithms are shape preservation and topology preservation. Most of the existing scaling algorithms focus on shape preservation and some pay attention to topology preservation. In this paper we propose a shrinking algorithm for binary images to preserve topology. The proposed algorithm shrinks the image by a factor of powers of 2 through iterative process. It divides the image into disjoint 2×2 rectangle blocks at each iterative step. Each block is in smooth region or in edge region. For smooth region, the block has the same color, so the corresponding pixel takes the same color as the block. For edge region, the block has different colors, so the corresponding pixel is processed according to human visual system and connectivity classification. Experimental results show that the proposed shrinking algorithm has better topology preservation and similar shape preservation in comparison with existing shrinking algorithms.