Classified Attribute Cluster Network Applied on Fractal Block Coding

摘要

The main drawback of Fractal Block Coding is its long time-consuming, and much work has been done for it. In this work, we advise a new clustering method, Classified Attribute Cluster Network (CACN). First, image blocks are classified according to their edge feature; then we apply Attribute Cluster Network on each sub-class. Experiments indicate, with achieving the similar image quality and compression ratio, CACN costs less encoding time and clustering time, especially for the larger domain pool.