为了保持图像的特征和细节，文章利用图像的方向导数局部地调整正则系数。
To preserve image features and fine part, the regular coefficient is locally adjusted according to the directional derivatives of the image.
图像融合过程采用了“选择”和“加权平均”混合的系数选择规则，在充分保留显著图像细节信息的前提下，进一步增强了算法的稳定性。
The combined rule of "selection" and "weighted averaging" is employed, which both preserves the source images' details and improves the algorithm's stability.
借助热红外图像再现实验和图像相关系数算法，证明该方法能较好地再现目标以及背景的热红外辐射特征和细节。
It is verified that this method could reproduce the properties and details of both targets and backgrounds by the thermal IR reproduction experiment and the algorithm of image correlation coefficient.
应用推荐