Multimodal biomedical image fusion: Use of log-Gabor and guided filters with non-subsampled contourlet transform

In this paper, we propose a multimodal biomedical image fusion approach based on log-Gabor and guided filters with non-sub sampled contourlet transform (NSCT). Here, first the source images are decomposed into low and high frequency sub-bands using NSCT. Experimentally it is proved that Gabor filter provides an effective encoding of natural images. Log-Gabor functions overcomes the bandwidth limitation of Gabor functions as this filter can be constructed with arbitrary bandwidth and optimized bandwidth can be used to produce a filter with minimal spatial extent. Due to this, log-Gabor filter is used to extract the meaningful low frequency details from the low frequency sub-bands. Similarly, literature shows that the guided filter is the edge preserving smoothing filter. In the proposed fusion approach we use the same to extract the high frequency details from the high frequency sub-bands. The local regional variance between the filtered outputs of both the source images is calculated to fuse the low and high frequency sub-bands. The final fused image is obtained by taking the inverse contourlet transform. The potential of the fusion method is evaluated by conducting the experiments on real medical images. Also, the results of the proposed fusion method are compared with the other state of the art methods. In addition to this, the different quantitative measures are also calculated to show the effectiveness of the proposed fusion method. Comparison of these results shows that the proposed fusion scheme with log-Gabor filter and guided filter results in better fusion of low and high frequency details when compared to the that with the other existing fusion approaches.