Visible and thermal image fusion using curvelet transform and brain storm optimization

In this paper, we propose a brain storm optimized image fusion framework in the curvelet transform domain that combines thermal image with the visual image to obtain a single informative fused image. The source images are decomposed using the curvelet transform and the high frequency sub-band coefficients are fused by maximum selection rule, whereas the low frequency sub-band coefficients are fused by weighted linear combination rule. The human intelligence based brain storm optimization (BSO) algorithm is employed to find the optimal weights in fusing low frequency sub-band coefficients. Simulation have been performed to compare our results with other multi resolution fusion methods such as gradient (GRAD) pyramid, shift invariant discrete wavelet transform (SIDWT) and non sub-sampled contourlet transform (NSCT). The quality of the fused image is assessed using five different quality metrics and the results indicate that the proposed method outperforms other multiresolution fusion methods in terms of both subjective and objective quality metrics.