A simple and efficient image fusion algorithm based on standard deviation in wavelet domain

Multimodal neurological image fusion based on adaptive biological inspired neural model in nonsubsampled Shearlet domain

“…Method 4: Fusion scheme based on the STD in WT domain by decomposing the each one of the reference images into 8 × 8 block.…”

“…Four image pairs out of complete CT‐MR data set is shown in Figure A,B, respectively. For presenting the comparative visual performance of the fused images, some of the methods developed previously are considered as fusion method (FM‐1), FM‐2, FM‐3, FM‐4, (mentioned above as method 4, Method 5, Method 7, and Method 9) respectively, FM‐5, and proposed MMIF method. Their fusion results are shown in Figure C‐H, respectively, and by observing these fused images, it is visualized that the resultant fused images obtained by the proposed MMIF approach have better visual ability along with better contrast and edge information that is supported by the quantitative value of the En, STD, and XEI parameters.…”

“…Yang et al introduced another WT‐based approach using different rules based on visibility and variance to fuse the different subcomponents . The authors proposed another approach based on STD using the decomposition of source images in the WT domain . The MMIF methods based on the WT are capable to reflect only 1‐D singularity and have limited number of directional information that leads to artifacts along the hf details and loss of the most important diagnostic information also.…”

“…9 The authors proposed another approach based on STD using the decomposition of source images in the WT domain. 10 The MMIF methods based on the WT are capable to reflect only 1-D singularity and have limited number of directional information that leads to artifacts along the hf details and loss of the most important diagnostic information also. Similar to the WT decomposition, another fusion approach based on singular value decomposition has been reported that worked better for multifocus image data, however, it does not work well for medical images.…”

Multistage multimodal medical image fusion model using feature‐adaptive pulse coupled neural network

“…images show more detail information along with the color preservation compared to the source MR-T2 and 99mTc-SPECT images. Furthermore, to explore the fusion performance of the HMMIF approach, a comparison is shown in Figures 5-8, wherein the fusion results of the proposed model are compared to V-2011, 5 N-2016, 7 GK-2016, 25 Y-2017 16 and Z-2018 28 fusion approaches. Figures 5-8A-E 25 shows good information and edge preservation from the source MR images, however, it suffers from poor spectral contrast and consistency hence shows over-brightness in some regions.…”