Multi-image fusion: optimal decomposition strategy with heuristic-assisted non-subsampled shearlet transform for multimodal image fusion
Jampani Ravi,B. V. Subbayamma,P. Vijaya Kumar,Yadavalli S. S. Sriramam,S. Marlin,Adlin Sheeba,N. V. Phani Sai Kumar
DOI: https://doi.org/10.1007/s11760-023-02906-3
IF: 1.583
2024-02-09
Signal Image and Video Processing
Abstract:Image fusion is significant in various distinct sectors of image processing, from remote sensing to medical applications. In recent years, real-valued wavelet transforms have been utilized to fuse images. This approach has offered enhancements against various poor approaches; however, this task lacks the shift variance and suffers from the directionality connected with its wavelet bases. Moreover, the conventional architecture of this hard wavelet decomposition implements a very hard resolution of filters to attain an essential quarter shift in the coefficient result. The establishment of image fusion methodology is to provide data integrated from distinct images to avoid inconsistency and redundancy presented among the images. This approach is utilized to enhance the utilization, reliability, accuracy, and interpretation of the data with the development of image data transparency by creating an accurate and clear detail of the monitored target. In this research work, a transform-aided image fusion mechanism is utilized to enhance the effectiveness in a better way. With the support of this approach, good "peak signal-to-noise ratio (PSNR)" with a minimum "mean square error (MSE)" can be achieved. Therefore, this work is aimed to implement a new multi-image fusion approach by fusing the normal images. Initially, the standard normal images are manually collected for the approach. Then, the decomposition of two images in the same scene is done through "optimal non-subsampled shearlet transform (ONSST)," where the attributes of NSST are optimized with the help of recommended fitness improved puzzle optimization algorithm (FIPOA). Moreover, the high-frequency fusion is done by optimal weighted average fusion, and low-frequency fusion is carried out by filter mapping-based fusion. In the end, the inverse ONSST is taken to get the final integrated images. The experimental analysis of the recommended approach is evaluated with various performance measures. The validation shows that the developed model attains 44.7%, 6.18%, 17.4%, 17.4%, and 9.7% enhanced performance than DOX-ONSST, AOA-ONSST, SFO-ONSST, and POA-ONSST in terms of standard deviation. The experimental analysis of the developed model shows better performance rather than the existing approaches. The image fusion is widely applicable in the field of clinical and healthcare applications.
engineering, electrical & electronic,imaging science & photographic technology