研究目的
To propose two new convex variational models for fusing noisy source images by utilizing fractional-order derivatives to represent image features and applying an alternating direction method of multiplier (ADMM) to solve the minimization problems in the proposed models.
研究成果
The proposed TVL1 and TVL2 methods effectively fuse and denoise noisy source images by utilizing fractional-order derivatives and TV regularization. Numerical experiments show that these methods outperform existing methods in terms of visual and quantitative measures. The TVL2 method, in particular, provides better fused images without undesired artifacts.
研究不足
The proposed models rely on manually tuning the regularization parameters, which may affect the fusion results. An adaptive parameter selection method could be considered in future work.
1：Experimental Design and Method Selection:
The study utilizes fractional-order derivatives to represent image features and proposes two new convex variational models for fusing noisy source images. The ADMM algorithm is applied to solve the minimization problems in the proposed models.
2：Sample Selection and Data Sources:
The study uses simulated and real images, including multi-focus and multi-modal images, corrupted by Gaussian noise with different levels.
3：List of Experimental Equipment and Materials:
The experiments are performed using Matlab (R2015b) on a desktop with a 3.40GHz Intel Core i3-2130 CPU and 4G RAM.
4：40GHz Intel Core i3-2130 CPU and 4G RAM.
Experimental Procedures and Operational Workflow:
4. Experimental Procedures and Operational Workflow: The proposed models are compared with existing methods (WTV and FSG) in terms of visual and quantitative measures (QMI and QP). The influence of model parameters (α, β, γ) on the fusion results is also tested.
5：Data Analysis Methods:
The performance of the proposed methods is evaluated using mutual information metric (QMI) and phase congruency metric (QP).
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
