| Abstrakt: |
Image fusion due to its cost-effectiveness and applicability in a broader range of applications makes it an emerging area in research. However, it is seen from the literature that most of the existing fusion algorithms are application-specific. As a result, the results obtained for different applications are limited. So, in this work, we propose an effective algorithm for better outcomes for different applications. For this, an adaptive image decomposition tool known as Hilbert vibration decomposition (HVD) is used. HVD decomposes an image into instantaneous energy components having amplitudes (image amplitudes) and frequencies. Unlike traditional multi-scale decomposition, the adaptive decomposition strategy used by HVD does not require any fixed cut-off frequency or pre-defined function basis and offers better spatial resolution. Then, we compute instantaneous detailed image amplitudes that generally contain significant information. These are obtained by subtracting the instantaneous image amplitudes from the source images. Further, we find the optimized weights with the help of a statistical approach, i.e., by using unbiased estimates and eigenvalues related to these instantaneous detailed image amplitudes. After this computation, the optimized weights are integrated with source images to generate the final fused image. The simulation of the proposed work is carried out using MATLAB software for multi-focus, medical, and visible-infrared image samples and compared with existing methods. It is seen that in comparison to traditional and some deep learning-based fusion works, the proposed work provides better/comparative outputs qualitatively as well as quantitatively. Moreover, the execution time of the proposed algorithm is comparable to that of other recent methods. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink