Image fusion algorithm based on fractal dimension and contrast in multi-wavelet transform domain

In order to make full use of texture features of image when fusing images, and taking into account the inherent advantages of fractal theory in this respect, a novel image fusion algorithm, which used fractal dimension and directional contrast, based on multi-wavelet transform was proposed in this paper. This fusion algorithm decomposed original images needed to be fused by multi-wavelet transform firstly, then fractal dimensions of the corresponding decomposing low-frequency coefficients were calculated by Differential Box Counting method, and the low frequency fusion rule based on fractal dimension was set up. The high frequency parts were fused by selecting method or weighted averaging method according to the directional contrast values. The fusing experiment using this algorithm to IR image and visible light image was done. In order to illustrate the performance of the algorithm, particularly the impact of introduction of fractal theory on the fusion performance, the result of those fusion algorithms that only uses multi-wavelets transform were given as well. Moreover, the objective indexes, which are image entropy, standard deviation and quality measure, were adopted to evaluate the comparative results of evaluating fusion quality. The results show that the leading of fractal dimension into image fusion processing is effective, and it performs obviously better in increasing the amount of information, the standard deviation and the quality of fused image.