A novel algorithm for multifocus image fusion based on contourlet Hidden Markov Tree model

Abstract

According to features of multifocus images and statistical characteristics of contourlet coefficients, a novel algorithm for multifocus image fusion based on contourlet Hidden Markov Tree model (con-HMT) is proposed. Multifocus images are used all together to train the contourlet HMT model. Then a new fusion rule for the high frequency is built. In this rule, the probability of a detailed coefficient corresponding to image edge, calculated directly from the HMT model, is chosen as the salience measure. Experimental results show that, for multifocus image fusion, the proposed algorithm provides more satisfying fusion results in terms of visual effect and objective evaluations, which proves its feasibility and validity.