MRI-SPECT image fusion integrates anatomical structures and functional information, yielding a unified and enriched representation. Medical image fusion techniques based on joint sparse representation (JSR) combine the complimentary qualities of source images using a single dictionary. However, these methods are not entirely effective in accurately representing the unique qualities of different modal signals. To solve this constraint, this study introduces a unique technique called coupled dictionary-based fusion (CDF), which successfully maintains geometrical properties from input images. In this method, the energy of gradients (EOG) serves as the primary metric for measuring activity levels, aiding in the selection of focus pairs from the MRI-SPECT image sets. Subsequently, an adaptive coupled dictionary is formed through an online dictionary learning (ODL) algorithm, which effectively exposes the inherent structures within the input images. By utilizing this trained dictionary, sparse coefficients are computed using the Lasso algorithm. To capture the key features of the input images, a novel fusion method based on higher valued sparse coefficients is developed. An experimental assessment including 12 pairs of MRI-SPECT image reveals the proposed CDF technique's computing efficiency and improved performance, both in terms of visual quality and statistical metrics, outperforming previous approaches.
MRI-SPECT Image Fusion with Coupled Sparse Dictionary
2023-11-22
1025204 byte
Conference paper
Electronic Resource
English
Sparse Fusion Based on SAM Elective Sample Dictionary Establishment
British Library Conference Proceedings | 2016
|Image Deblurring with Coupled Dictionary Learning
British Library Online Contents | 2015
|Image denoising via sparse representation using rotational dictionary
British Library Online Contents | 2014
|Sparse Representation Based Fisher Discrimination Dictionary Learning for Image Classification
British Library Online Contents | 2014
|Image fusion with nonsubsampled contourlet transform and sparse representation
British Library Online Contents | 2013
|