An integrated classifier in classified coding

Block coding is one of the most common schemes in image data compression. To improve the performance of block coding, a classification can be applied to the blocks prior to coding. This results in an adaptive block coding method, i.e., classified coding. In this paper, an integrated classifier taking the features of the human vision system (HVS) into account in classified coding is proposed. First, we present a classifier based on the local contrast sensitivity of the HVS. Compared with the commonly used local variance-based classifier (LVC), it possesses lower computational complexity while preserving almost the same performance. To improve upon the single-parameter classifiers´ poor adaptivity, we, then, propose an integrated classifier which is composed of three independent classification units. It exhibits the complementarity of different classifiers. The simulation results demonstrate that the proposed classifier performs better than the LVC