A New Approach for Low-resolution Image Registration Based on Weighted MI and Powell

2018 2nd International Conference on Artificial Intelligence Applications and Technologies

计算机科学

Zhang, Shuping^1

College of Engineering, Shanghai Polytechnic University, Shanghai

201209, China^1

关键词: Low resolution;    Low resolution images;    Mutual informations;    Optimizers;    Registration scheme;    Similarity measure;    Strong robustness;    Weight coefficients;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/435/1/012008/pdf
DOI  :  10.1088/1757-899X/435/1/012008

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

Mutual information (MI) has been successfully applied to the registration of medical imagery and SR reconstruction because of its robustness. However, it also has obvious drawbacks that it is susceptible to local minima. For this purpose, an image registration scheme based on weighted mutual information is proposed in this paper. By introducing a reasonable weight coefficient, the gray-levels can exert greater impact on computation of entropy. And then, the weighted mutual information (WMI) which is designed as a new similarity measure is exploited by the Powell optimizer to refine the preregistration results. The experimental results suggest that the proposed algorithm has strong robustness and adaptive capacity for noise, and is convergent and efficient in resolving the given test problems.