Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
6902329 | Procedia Computer Science | 2017 | 8 Pages |
Abstract
Multiscale Transforms (MST) are widely used in fusing multimodal images, however these suffer from drawbacks like poor contrast, low edge detection, blurring, redundancy and high execution time. Besides, choice of decomposition levels and fusion rules is also a challenge. Sparse Representation (SR) based image fusion techniques overcome these drawbacks. This paper aims the implementation of MST and MST-SR based fusion techniques for multimodal and multiresolution brain images. A novel technique based on LP-SR is proposed. Both subjective and objective evaluation is made on multiple sets of source images. LP-SR show superior results for contrast, SSIM and UIQI.
Related Topics
Physical Sciences and Engineering
Computer Science
Computer Science (General)
Authors
Leena Chandrashekar, A. Sreedevi,